merge refactors pr #364 from @michilyy

* Save to DB only unique trackhashes * Add check if track already exists in playlist * replace all paths with `pathlib.Path` * `architecture.md`: * add config folder layout `config.py`: * fix bug where `pathlib.Path` cannot be serialized `files.py`: * remove unused imports * update path concatenation to `pathlib.Path` * add config-folder creation `imgserver.py`: * fix serialisation bug `playlistlib.py`: * update path concatenation to `pathlib.Path` * update all `settings.Paths` usages to new singleton `Paths` class. * update all usages of `settings.Paths` * `files.py`: * rework assets copy function. * remove unused loop and unused `shutil.copy2` function `settings.py` * fix recursion exception in `Paths` * `settings.py`: * remove Singleton and `@property` todos from `Paths` * `__init__.py`: * remove now unused function `create_config_dir()` `setup.files`: * remove because merged into `settings.Paths()` for more central and clear flow how the base path gets decided `settings.py`: * add `copy_assets` function `start_swingmusic.py`: * add configurable settings.Paths class `__main__.py`: * update click to used correct default path * remove wrong commited egg files * remove change in the wrong branch * add forgotten `property` decorator update `get_files_and_dirs` to use pathlib where possible `config.py`: * update type annotation `folders.py`: * convert `pathlib` to posix path where needed for sub-functions `folderlib.py`: * rework `get_files_and_dirs` to use `pathlib` where possible `settings.py`: * add forgotten `@property` `start_swingmusic.py`: * remove second `log_startup_info()` * `artistlib.py`: * fix calling property `tagger.py`: * fix comparing elements in `pathlib.Path` * add support for repeating lyrics. * rework lyrics api and lib * update most path functions. add type-hint pathlib where needed * for serialization paths are converted to posix path * use `open` instead of `os.open` update `metaclass` with constant * fix initial config exception if empty file existed * update `userConfig` with `InitVar` to be excluded from `asdict` * remove `is_windows_slash()` rework path function to use pathlib * convert `pathlib.Path` to `str` for serialization * fixing bug with str + pathlib * `__main__.py`: * update click to use package version * remove now unused function `print_version` `filesystem.py`: * rework `CWD` to use importlib `pyproject.toml`: * disable namespace for `importlib.resources` to work correctly * update `lyrics.py`: * remove unused functions * simplify functions * fix bug where assets get created on root * remove unused code * update lyrics for clearer structure. * add support for unsynced lyrics * fix wrong return type in unsynced lyrics * update `/check` to use `send_lyrics` * prefer tags to duplicates * `lyrics.py`: * add docs to a function group * `logger.py`: * add logging config dict. * combine Logging into one file * add socket logger * add debug mode to logger * add JSONL formater * `logger.py`: * update config to directly use the formater. resolves circular import exception `__main__.py`: * add logger setup to main `start_swingmusic.py`: * add debug option to cli * `lyrics.py`: * add offset support * add `setuptools-scm` to get version from git * add support for docker build with scm * add support for docker build with scm need someone who can test the changes workflow * update all usage of `version.txt` to `metadata.version()` * 2x update all usage of `version.txt` to `metadata.version()` * update to no local_scheme version * provide fix for #331. convert `sql.Row` and `TrackTable` to dict before converting to dataclass. * fix `__main__.py`: * wrong import and uncommited changes * add debug and base_path parameter * fix logger pathlib * add client build workflow * set name * split client from build * try fixing builds * try another fix * try also another fix * try again something new * try again something new * change runner * fix failed run because of malformed runner * add wheel builds * remove systems from pure python build * add isolated pyinstaller build * artifacts with names * wrong wheel path * try fetch-depth for tag fetch * disable fail-fast. add wheel installation * add install system packages * add debug * fix wheel install fix pyinstaller spec file * try fix for pyinstaller * try another fix * build on release * add concrete release types * only run on released or pre-released * try release upload * reformat upload * fix needs tag * identifiable pyinstaller builds * compress client folder before uploading * update to src build * remove no more needed aarch64 build script rename pyinstaller assets to lowercase * remove unneeded code * fix: save to DB only unique track hashes * replace click with argparse * set concrete types in argparse * replace manuall path usages with pathlib * remove unused `configs.py` file * reformat `start_swingmusic.py` * fix empty set startup exception * optimizing static files serve function * fixing bug in optimisation of static files serve function * fix folder view bug * colorlib.py: * fix wrong type exception * remove singe use Index_everything class * update logging of populate.py * cleanup files * fix settings.py Paths copy function. Created folder on file. * add exist check to folder * remove unused `INFO` class * fix multiprocessing bug on windows * potential icon fix for pyinstaller fix multiple logging bug * fix argparse config path bug add jobs file * cleanup code fragments fix logging issue add notes to function * note that concurrent creates own sys.modules * refactor some lyrics plugin condition remove unused import from hashing * refactor taglib.py * update import statements to be static * playlistlib.py: * refactoring and more doc strings populate.py: * add poc bugfix settings.py: * add typehint * possible bugfix for multitreading globals * folder.py: * add check if provided path is absolute populate.py: * add bug note settings.py: * add possible error from Singleton implementation start_swingmusic.py: * correct spelling * pass resolved path to Paths tagger.py: * add logging * trying out fixes for multithreading * only upload results not metadata * fix build action again * folder.py: * strictly use pathlib where possible folderlib.py: * add missing docstring to function, who really need it. track.py: * refactor some code folder.py: * refactor some more code * Merge DBPath class and Paths class. Update all usages of DBPath folderslib.py: * fix bug with logging taglib.py: * add missing docstring settings.py: * merge classes * refactor * network.py: * add more docstring config.py: * update pathlib usage tools.py: * refactor * add docstrings * colorlib.py: * add docstring Refactor App builder into grouped config settings. * update assets access for migration * Update FUNDING.yml * Update FUNDING.yml * upgrade tinytag in requirements.txt * update readme * update license * update readme * Update README.md * Update README.md * cleanup requirements.txt remove unused import in audio_segment.py add entrypoint.sh for appimage support update pyproject.toml for optional dependencies add appimage to github workflow * fix invalid workflow file * AppImage build needs more research. Commenting for now * testing a new build workflow * add libev installation * update workflow to new optional dependencies * trying again another fix * finally fix all optional deps installation correctly * remove AppImage poc * albumslib.py: * add docstring folder.py: * add unix path fix update logger name to `__name__` * update build with docker update Dockerfile with git fix typo in lyrics.py add dynamic deps back * add log for static folder * add missing import * add some more todos * add support for AppImages even when it's not perfect. * quick bugfix for wrong appimage config path * fix uploading not finding AppImages builds aka wrong pattern * optimise docker build by using artifacts. Add client path option. change docstring to sphinx format * add todos * Now support AppImages for real: manually build AppImage as we are building a complex project. * fix missing dep in AppImage build * add full AppImage metadata * add missing image file. * only update swingmusic appimage not tool * add todo and fix AppImage build again. * Try fixing some path mixup in AppImage build * add debug tag to action * correct path to appimage folder * do not download tool before checkout * Another fix for path in appimage build * extend config files with more information * default client dir is now inside the config dir. TODOs updated. * default client dir is now inside the config dir. TODOs updated. Add priority todos. * Auto download client when client not found. Respects user provided dir. * rename `requests` submodule to `request` * poc for arm AppImage builds * try out another fix * fix typo in build.yml * add missing arch tag * fix uploading double names * unique naming * enable fallback version for project. * do not download client into readonly dir. * fix relative client download path. Client was resolved into parent of config. * remove client backup path as client is now downloadable * `Paths` checks if config folder exists and creates it if necessary. logger no more creates the config folder. `app_builder.py`: static route no more with '/client' * path are only created in MainProcess. fix gz file not found. * move assets into src and update usages accordingly * remove solved todos * Only upload artefacts if not draft/master aka only on tag * wrong type in assets copy * update log with correct priority * add debug statements and logging to Paths * remove debugging statement * remove double version tag from docker build * fork save release protection * fix typo * add fallback client dir for static builds. * update argparse to new param * add missing import pathlib * add sparse checkout as we do not need everything downloaded * add assets copy check * init logger bevor Paths * remove unused import * check if logdir exists and create if not * only add exec info to file * remove exception log from cli * move logging into main. Allows tools support again. * UserConfig now correctly uses _finished key. Bug where _finished was never written * double save serverId. update root_dir to trow no exception on init. remove debug param * clean up TODOs --------- Co-authored-by: skilletfun <skilletfun.laptew.sergey@yandex.ru> Co-authored-by: Mungai Njoroge <geoffreymungai45@gmail.com>
2026-06-05 04:53:01 +00:00 · 2025-08-28 09:28:11 +00:00
parent b4b0a6e11f
commit e770606567
197 changed files with 2961 additions and 2150 deletions
@@ -0,0 +1,3 @@
+"""
+This module contains all the data processing and non-API libraries
+"""
@@ -0,0 +1,30 @@
+"""
+Contains methods relating to albums.
+"""
+
+from swingmusic.models.track import Track
+
+
+def remove_duplicate_on_merge_versions(tracks: list[Track]):
+    """
+    Removes duplicate tracks when merging versions of the same album.
+    """
+    # TODO!
+    pass
+
+
+def sort_by_track_no(tracks: list[Track]) -> list[Track]:
+    """
+    Sort tracks by track number.
+    Track numbers cannot be longer than three positions.
+
+    :param tracks: List of Tracks
+    :return: Sorted list of Tracks
+    """
+    for t in tracks:
+        track = str(t.track).zfill(3)
+        t._pos = int(f"{t.disc}{track}")
+
+    tracks = sorted(tracks, key=lambda t: t._pos)
+
+    return tracks
@@ -0,0 +1,187 @@
+import os
+import time
+import random
+import urllib
+import requests
+
+from io import BytesIO
+from pathlib import Path
+from concurrent.futures import ProcessPoolExecutor
+
+from PIL import Image, PngImagePlugin, UnidentifiedImageError
+from requests.exceptions import ConnectionError as RequestConnectionError
+from requests.exceptions import ReadTimeout
+
+from swingmusic import settings
+from swingmusic.models.artist import Artist
+from swingmusic.store.artists import ArtistStore
+
+# from swingmusic.db.libdata import ArtistTable
+
+# from swingmusic.store import artists as artist_store
+# from swingmusic.store.tracks import TrackStore
+from swingmusic.utils.hashing import create_hash
+from swingmusic.utils.progressbar import tqdm
+
+
+LARGE_ENOUGH_NUMBER = 100
+PngImagePlugin.MAX_TEXT_CHUNK = LARGE_ENOUGH_NUMBER * (1024**2)
+# https://stackoverflow.com/a/61466412
+
+
+def get_artist_image_link(artist: str):
+    """
+    Returns an artist image url.
+    """
+    response: requests.Response | None = None
+
+    def make_request():
+        query = urllib.parse.quote(artist)  # type: ignore
+        url = f"https://api.deezer.com/search/artist?q={query}"
+        user_agents = [
+            "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36",
+            "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/14.0 Safari/605.1.15",
+            "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/92.0.4515.107 Safari/537.36",
+            "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:90.0) Gecko/20100101 Firefox/90.0",
+            "Mozilla/5.0 (iPhone; CPU iPhone OS 14_6 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/14.0 Mobile/15E148 Safari/604.1",
+        ]
+        headers = {
+            "User-Agent": random.choice(user_agents),
+            "Accept": "application/json, text/plain, */*",
+            "Accept-Language": "en-US,en;q=0.9",
+            "Referer": "https://www.deezer.com/",
+            "Origin": "https://www.deezer.com",
+        }
+        return requests.get(url, headers=headers, timeout=30)
+
+    for attempt in range(5):
+        try:
+            response = make_request()
+            try:
+                data = response.json()
+            except requests.exceptions.JSONDecodeError:
+                return None
+
+            for res in data["data"]:
+                res_hash = create_hash(res["name"], decode=True)
+                artist_hash = create_hash(artist, decode=True)
+
+                if res_hash == artist_hash:
+                    return str(res["picture_big"])
+
+            return None
+        except (RequestConnectionError, ReadTimeout, IndexError, KeyError):
+            if attempt == 4:
+                print("Failed to get artist image link  ")
+
+            if attempt <= 4:
+                time.sleep(10)
+            else:
+                return None
+
+        # except (IndexError, KeyError):
+        #     print(f"Encountered index/key error in attempt {attempt}")
+        #     if response is not None:
+        #         print(response.headers)
+
+        #     return None
+
+
+# TODO: Move network calls to utils/network.py
+class DownloadImage:
+    def __init__(self, url: str, name: str) -> None:
+        img = self.download(url)
+
+        if img is None:
+            return
+
+        sm_path = settings.Paths().sm_artist_img_path / name
+        lg_path = settings.Paths().lg_artist_img_path / name
+        md_path = settings.Paths().md_artist_img_path / name
+
+        entries = [
+            (lg_path, None),  # save in the original size
+            (sm_path, settings.Defaults.SM_ARTIST_IMG_SIZE),
+            (md_path, settings.Defaults.MD_ARTIST_IMG_SIZE),
+        ]
+
+        self.save_img(img, entries)
+
+    @staticmethod
+    def download(url: str) -> Image.Image | None:
+        """
+        Downloads the image from the url.
+        Retries after 10 seconds on a connection error.
+        """
+        for attempt in range(2):
+            try:
+                response = requests.get(url, timeout=10)
+                return Image.open(BytesIO(response.content))
+            except (RequestConnectionError, requests.Timeout, ReadTimeout):
+                if attempt == 0:
+                    time.sleep(10)
+                else:
+                    return None
+            except UnidentifiedImageError:
+                return None
+
+    @staticmethod
+    def save_img(img: Image.Image, entries: list[tuple[Path, int | None]]):
+        """
+        Saves the image to the destinations.
+        """
+        ratio = img.width / img.height
+        for entry in entries:
+            path, size = entry
+
+            if size is None:
+                img.save(path, format="webp")
+                continue
+
+            img.resize((size, int(size / ratio)), Image.Resampling.LANCZOS).save(
+                path, format="webp"
+            )
+
+
+class CheckArtistImages:
+    def __init__(self):
+        # read all files in the artist image folder
+        storeArtists = ArtistStore.get_flat_list()
+        path = settings.Paths().sm_artist_img_path
+        processed = set(i.replace(".webp", "") for i in os.listdir(path))
+
+        unprocessed = (
+            artist for artist in storeArtists if artist.artisthash not in processed
+        )
+
+        num_workers = max(1, (os.cpu_count() or 1) // 2)
+
+        with ProcessPoolExecutor(max_workers=num_workers) as executor:
+            res = list(
+                tqdm(
+                    executor.map(self.download_image, unprocessed),
+                    total=len(storeArtists) - len(processed),
+                    desc="Downloading missing artist images",
+                )
+            )
+
+            list(res)
+
+    @staticmethod
+    def download_image(artist: Artist):
+        """
+        Checks if an artist image exists and downloads it if not.
+
+        :param artist: The artist name
+        """
+        img_path = (
+                settings.Paths().sm_artist_img_path / f"{artist.artisthash}.webp"
+        )
+
+        if img_path.exists():
+            return
+
+        url = get_artist_image_link(artist.name)
+
+        if url is not None:
+            return DownloadImage(url, name=f"{artist.artisthash}.webp")
@@ -0,0 +1,286 @@
+"""
+Contains everything that deals with image colour extraction.
+"""
+import logging
+import os
+import pathlib
+
+import colorgram
+from pathlib import Path
+from typing import Generator
+from swingmusic.utils.progressbar import tqdm
+from concurrent.futures import ProcessPoolExecutor, as_completed
+
+from swingmusic import settings
+from swingmusic.store.albums import AlbumStore
+from swingmusic.db.userdata import LibDataTable
+from swingmusic.store.artists import ArtistStore
+
+log = logging.getLogger(__name__)
+
+def get_image_colors(image: pathlib.Path, count=1) -> list[str]:
+    """
+    Extracts ``count`` numbers of the most dominant colours from an image.
+
+    :params image: Path to image.
+    :params count: How many colours should be extracted?
+    :returns: List["rgb(red, green, blue)", ...]
+    """
+
+    if image.exists():
+        colors = sorted(colorgram.extract(image, count), key=lambda c: c.hsl.h)
+    else:
+        return []
+
+    formatted_colors = []
+
+    for color in colors:
+        color = f"rgb({color.rgb.r}, {color.rgb.g}, {color.rgb.b})"
+        formatted_colors.append(color)
+
+    return formatted_colors
+
+
+def process_color(item_hash: str, is_album=True) -> list[str]:
+    """
+    Parse colours from images associated with song
+
+    :param item_hash: hash of item for colour calculation
+    :param is_album: if item is an album
+    :return: list with colour strings
+    """
+
+    if is_album:
+        path = settings.Paths().sm_thumb_path
+    else:
+        path = settings.Paths().sm_artist_img_path
+
+    path = path / (item_hash + ".webp")
+
+    if not path.exists():
+        return []
+
+    return get_image_colors(path)
+
+
+def extract_color_worker(item_data: dict) -> dict:
+    """
+    Generic worker function for extracting colours in parallel.
+    Returns data to main process for batch database operations.
+    Works for both albums and artists based on item_data configuration.
+    """
+    hash_field: str = item_data["hash_field"]
+    path_func: Path = item_data["path_func"]
+    item_hash: str = item_data[hash_field]
+
+    path = path_func / (item_hash + ".webp")
+
+    if not path.exists():
+        return {hash_field: item_hash, "color": None, "error": "Image not found"}
+
+    colors = get_image_colors(path)
+
+    if not colors:
+        return {
+            hash_field: item_hash,
+            "color": None,
+            "error": "Color extraction failed",
+        }
+
+    return {hash_field: item_hash, "color": colors[0], "error": None}
+
+
+class ColorProcessor:
+    """
+    Generic color processor for extracting dominant colors from images.
+    Uses multiprocessing for parallel color extraction and batch database operations.
+    """
+
+    def __init__(
+        self,
+        item_type: str,
+        store: AlbumStore | ArtistStore,
+        path_func: Path,
+        hash_field: str,
+    ):
+        """
+        Initialize the color processor.
+
+        Args:
+            item_type: Type of item ("album" or "artist")
+            store: Store object (AlbumStore or ArtistStore)
+            path_func: Function to get the image path
+            hash_field: Name of the hash field ("albumhash" or "artisthash")
+        """
+        self.item_type = item_type
+        self.store = store
+        self.path_func = path_func
+        self.hash_field = hash_field
+
+        # Read existing colors from database to filter out already processed items
+        existing_colors = set()
+        for color_data in LibDataTable.get_all_colors(item_type):
+            if color_data["color"]:
+                existing_colors.add(color_data["itemhash"])
+
+        # Filter items that need color processing
+        items_needing_colors = self._get_items_needing_colors(existing_colors)
+
+        if not items_needing_colors:
+            return
+
+        self._process_colors_parallel(items_needing_colors)
+
+    def _get_items_needing_colors(
+        self, existing_colors: set
+    ) -> Generator[dict, None, None]:
+        """
+        Generator that yields items needing color processing.
+        """
+        for item in self.store.get_flat_list():
+            # Skip if item already has color in memory store
+            if item.color:
+                continue
+
+            # Skip if item already has color in database
+            item_hash = getattr(item, self.hash_field)
+            if item_hash in existing_colors:
+                continue
+
+            yield {
+                self.hash_field: item_hash,
+                "item_type": self.item_type,
+                "path_func": self.path_func,
+                "hash_field": self.hash_field,
+            }
+
+    def _process_colors_parallel(self, items: Generator[dict, None, None]) -> None:
+        """
+        Process colors using multiprocessing and batch database operations.
+        """
+        items_list = list(items)
+
+        if not items_list:
+            return
+
+        cpus = max(1, (os.cpu_count() or 1) // 2)
+        batch_size = 20  # Process results in batches
+
+        with ProcessPoolExecutor(max_workers=cpus) as executor:
+            # Submit all jobs
+            future_to_item = {
+                executor.submit(extract_color_worker, item): item for item in items_list
+            }
+
+            batch = []
+            processed_count = 0
+
+            # Process results as they complete
+            progress_bar = tqdm(
+                as_completed(future_to_item),
+                total=len(items_list),
+                desc=f"Processing {self.item_type} colors",
+            )
+
+            for future in progress_bar:
+                try:
+                    result = future.result()
+
+                    if result["color"] is not None:
+                        batch.append(result)
+
+                    # Process batch when it reaches batch_size or we're done
+                    if len(batch) >= batch_size or processed_count + 1 >= len(
+                        items_list
+                    ):
+                        if batch:
+                            self._process_batch(batch)
+                            batch = []
+
+                    processed_count += 1
+
+                except Exception as e:
+                    item_data = future_to_item[future]
+                    item_hash = item_data[self.hash_field]
+                    log.error(f"Error processing {self.item_type} {item_hash}: {e}")
+
+    def _process_batch(self, batch: list[dict]) -> None:
+        """
+        Process a batch of color results - update database and memory stores.
+        """
+        if not batch:
+            return
+
+        # Prepare database records
+        db_inserts = []
+        db_updates = []
+
+        for result in batch:
+            item_hash = result[self.hash_field]
+            color = result["color"]
+
+            # Check if record exists in database
+            existing_record = LibDataTable.find_one(item_hash, type=self.item_type)
+
+            if existing_record is None:
+                db_inserts.append(
+                    {
+                        "itemhash": self.item_type + item_hash,
+                        "color": color,
+                        "itemtype": self.item_type,
+                    }
+                )
+            else:
+                db_updates.append(
+                    {"itemhash": self.item_type + item_hash, "color": color}
+                )
+
+        # Batch database operations
+        if db_inserts:
+            LibDataTable.insert_many(db_inserts)
+
+        if db_updates:
+            for update_data in db_updates:
+                clean_hash = update_data["itemhash"].replace(self.item_type, "")
+                LibDataTable.update_one(clean_hash, {"color": update_data["color"]})
+
+        # Update in-memory store
+        store_map = getattr(self.store, f"{self.item_type}map")
+
+        for result in batch:
+            item_hash = result[self.hash_field]
+            color = result["color"]
+
+            item = store_map.get(item_hash)
+            if item:
+                item.set_color(color)
+
+
+class ProcessAlbumColors:
+    """
+    Extracts the most dominant color from the album art and saves it to the database.
+    Uses multiprocessing for parallel color extraction and batch database operations.
+    """
+
+    def __init__(self) -> None:
+        ColorProcessor(
+            item_type="album",
+            store=AlbumStore,
+            path_func=settings.Paths().sm_thumb_path,
+            hash_field="albumhash",
+        )
+
+
+class ProcessArtistColors:
+    """
+    Extracts the most dominant colour from the artist art and saves it to the database.
+    Uses multiprocessing for parallel colour extraction and batch database operations.
+    """
+
+    def __init__(self) -> None:
+        ColorProcessor(
+            item_type="artist",
+            store=ArtistStore,
+            path_func=settings.Paths().sm_artist_img_path,
+            hash_field="artisthash",
+        )
@@ -0,0 +1,37 @@
+from typing import Any
+from swingmusic.store.albums import AlbumStore
+from swingmusic.store.artists import ArtistStore
+from swingmusic.store.tracks import TrackStore
+
+
+def get_extra_info(hash: str, type: str):
+    """
+    Generates extra info for a track, album or artist, which will be stored
+    in the database (in favorites, playlists and scrobble data) for backup and restore.
+
+    The extra info contains all the fields needed to reconstruct the itemhash. The track contains an additional filepath field which can be used to locate the file when restoring.
+    """
+    extra: dict[str, Any] = {}
+
+    if type == "track":
+        trackentry = TrackStore.trackhashmap.get(hash)
+        if trackentry is not None:
+            track = trackentry.get_best()
+
+            extra["filepath"] = track.filepath
+            extra["title"] = track.title
+            extra["artists"] = [a["name"] for a in track.artists]
+            extra["album"] = track.albumhash
+
+    elif type == "album":
+        album = AlbumStore.get_album_by_hash(hash)
+        if album is not None:
+            extra["albumartists"] = [a["name"] for a in album.albumartists]
+            extra["title"] = album.title
+
+    elif type == "artist":
+        artist = ArtistStore.get_artist_by_hash(hash)
+        if artist is not None:
+            extra["name"] = artist.name
+
+    return extra
@@ -0,0 +1,156 @@
+import pathlib
+from pathlib import Path
+import logging
+
+from swingmusic.lib.sortlib import sort_folders, sort_tracks
+from swingmusic.models import Folder
+from swingmusic.serializers.track import serialize_tracks
+from swingmusic.utils.filesystem import SUPPORTED_FILES
+from swingmusic.store.folder import FolderStore
+
+log = logging.getLogger(__name__)
+
+def create_folder(path: str, trackcount=0) -> Folder:
+    """
+    Creates a folder object from a path.
+    """
+    folder = Path(path)
+
+    return Folder(
+        name=folder.name,
+        path=folder.as_posix() + "/",
+        is_sym=folder.is_symlink(),
+        trackcount=trackcount,
+    )
+
+
+def get_folders(paths: list[str]):
+    """
+    Filters out folders that don't have any tracks and
+    returns a list of folder objects.
+    """
+    folders = FolderStore.count_tracks_containing_paths(paths)
+    return [
+        create_folder(f["path"], f["trackcount"])
+        for f in folders
+        if f["trackcount"] > 0
+    ]
+
+
+def get_files_and_dirs(
+    path: pathlib.Path,
+    start: int,
+    limit: int,
+    tracksortby: str,
+    foldersortby: str,
+    tracksort_reverse: bool,
+    foldersort_reverse: bool,
+    tracks_only: bool = False,
+    skip_empty_folders=True,
+) -> dict[str: list|int|str]:
+    """
+    Scan folder for files and folders.
+    Will only return files in `swingmusic.utils.filesystem.SUPPORTED_FILES`.
+    If `skip_empty_folders` is True
+
+    :param path:
+    :param start:
+    :param limit:
+    :param tracksortby:
+    :param foldersortby:
+    :param tracksort_reverse:
+    :param foldersort_reverse:
+    :param tracks_only: If True, will only return tracks with no folders
+    :param skip_empty_folders: If True, will call recursively and skip empty folders until >0 supported file found.
+    :returns: List of tracks and folders in that immediate path.
+    """
+
+    path = pathlib.Path(path)
+
+    # if file or non-existent
+    if not path.exists() or not path.is_dir():
+        return {
+            "path": path.as_posix(),
+            "tracks": [],
+            "folders": [],
+            "total": 0
+        }
+
+
+    # iter through all folders
+    # add files with supported suffix
+    # ignore hidden folder
+    dirs, files = [], []
+    for entry in path.iterdir():
+        ext = entry.suffix.lower()
+
+        if entry.is_dir() and not entry.stem.startswith("."):
+            dirs.append((entry / "").as_posix())
+            # only append as posix for FolderStore and sort_folder function
+            # TODO: rework everything to support pathlib
+            # add a trailing slash to the folder path
+            # to avoid matching a folder starting with the same name as the root path
+            # eg. .../Music and .../Music VideosI
+
+        elif entry.is_file() and ext in SUPPORTED_FILES:
+            files.append(entry)
+
+    """
+    # sort files by most recent
+    # TODO: rework if realy needed.
+    files_with_mtime = []
+    for file in files:
+        try:
+            files_with_mtime.append(
+                {
+                    "path": file.as_posix(),
+                    "time": file.lstat().st_mtime,
+                }
+            )
+        except OSError as e:
+            log.error(e)
+
+    files_with_mtime.sort(key=lambda f: f["time"])
+    files = [f["path"] for f in files_with_mtime]
+    """
+
+    # if supported files were found
+    # convert files to tracks
+    tracks = []
+    if len(files) > 0:
+        if limit == -1:
+            limit = len(files)
+
+        # only return tracks already indexed by us
+        tracks = list(FolderStore.get_tracks_by_filepaths(files))
+        tracks = sort_tracks(tracks, tracksortby, tracksort_reverse)
+        tracks = tracks[start : start + limit]
+
+
+    folders = []
+    if not tracks_only:
+        folders = get_folders(dirs)
+        folders = sort_folders(folders, foldersortby, foldersort_reverse)
+
+    if skip_empty_folders and len(folders) == 1 and len(tracks) == 0:
+        # INFO: When we only have one folder and no tracks,
+        # skip through empty folders.
+        # Call recursively with the first folder in the list.
+        return get_files_and_dirs(
+            folders[0].path,
+            start=start,
+            limit=limit,
+            tracksortby=tracksortby,
+            foldersortby=foldersortby,
+            tracksort_reverse=tracksort_reverse,
+            foldersort_reverse=foldersort_reverse,
+            tracks_only=tracks_only,
+            skip_empty_folders=True,
+        )
+
+    return {
+        "path": path.as_posix(),
+        "tracks": serialize_tracks(tracks),
+        "folders": folders,
+        "total": len(files),
+    }
@@ -0,0 +1,30 @@
+from swingmusic.db.userdata import MixTable
+from swingmusic.plugins.mixes import MixesPlugin
+
+
+def find_mix(mixid: str, sourcehash: str):
+    """
+    Find a mix in the homepage store or the db.
+    """
+    from swingmusic.store.homepage import HomepageStore
+
+    mixtype = "custom_mixes" if mixid[0] == "t" else "artist_mixes"
+
+    # INFO: Try getting the mix from the homepage store
+    mix = HomepageStore.get_mix(mixtype, mixid)
+    if mix and mix["sourcehash"] == sourcehash:
+        return mix
+
+    # INFO: Get the mix from the db
+    mix = MixTable.get_by_sourcehash(sourcehash)
+
+    if not mix:
+        return None
+
+    if mixtype == "custom_mixes":
+        mix = MixesPlugin.get_track_mix(mix)
+
+        if not mix:
+            return None
+
+    return mix.to_dict()
@@ -0,0 +1,170 @@
+import os
+import pathlib
+from swingmusic.db.userdata import PlaylistTable
+from swingmusic.lib.home import find_mix
+from swingmusic.lib.home.recentlyadded import get_recently_added_playlist
+from swingmusic.lib.home.recentlyplayed import get_recently_played_playlist
+from swingmusic.models.logger import TrackLog
+from swingmusic.store.albums import AlbumStore
+from swingmusic.store.artists import ArtistStore
+from swingmusic.store.tracks import TrackStore
+
+
+def create_items(entries: list[TrackLog], limit: int):
+    """
+    TODO: rework so that returns a dict with
+    {
+        "recently_played": ...,
+        "artist_mixes_for_you": ...
+    }
+    also keep in mind that the web-ui is beeing translated.
+    """
+    custom_playlists = [
+        {"name": "recentlyadded", "handler": get_recently_added_playlist},
+        {"name": "recentlyplayed", "handler": get_recently_played_playlist},
+    ]
+
+    items = []
+    added = set()
+
+    for entry in entries:
+        if len(items) >= limit:
+            break
+
+        if entry.source in added:
+            continue
+
+        added.add(entry.source)
+
+        if entry.type == "mix":
+            if not entry.type_src:
+                continue
+
+            splits = entry.type_src.split(".")
+
+            try:
+                mixid = splits[0]
+                sourcehash = splits[1]
+            except IndexError:
+                continue
+
+            # INFO: Get mix from homepage store
+            mix = find_mix(mixid, sourcehash)
+            if not mix:
+                continue
+
+            items.append(
+                {
+                    "type": "mix",
+                    "hash": entry.type_src,
+                    "timestamp": entry.timestamp,
+                }
+            )
+            continue
+
+        if entry.type == "album":
+            album = AlbumStore.albummap.get(entry.type_src)
+
+            if album is None:
+                continue
+
+            item = {
+                "type": "album",
+                "hash": entry.type_src,
+                "timestamp": entry.timestamp,
+            }
+
+            items.append(item)
+            continue
+
+        if entry.type == "artist":
+            artist = ArtistStore.artistmap.get(entry.type_src)
+
+            if artist is None:
+                continue
+
+            items.append(
+                {
+                    "type": "artist",
+                    "hash": entry.type_src,
+                    "timestamp": entry.timestamp,
+                }
+            )
+
+            continue
+
+        if entry.type == "folder":
+            folder = entry.type_src
+
+            if not folder:
+                continue
+
+            if not folder.endswith("/"):
+                folder += "/"
+
+            is_home_dir = entry.type_src == "$home"
+
+            if is_home_dir:
+                folder = os.path.expanduser("~")
+
+            if not pathlib.Path(folder).exists():
+                continue
+
+            item = {
+                "type": "folder",
+                "hash": folder,
+                "timestamp": entry.timestamp,
+            }
+
+            items.append(item)
+            continue
+
+        if entry.type == "playlist":
+            is_custom = entry.type_src in [i["name"] for i in custom_playlists]
+
+            if is_custom:
+                items.append(
+                    {
+                        "type": "playlist",
+                        "hash": entry.type_src,
+                        "timestamp": entry.timestamp,
+                        "is_custom": True,
+                    }
+                )
+                continue
+
+            playlist = PlaylistTable.get_by_id(entry.type_src)
+            if playlist is None:
+                continue
+
+            item = {
+                "type": "playlist",
+                "hash": entry.type_src,
+                "timestamp": entry.timestamp,
+            }
+
+            items.append(item)
+            continue
+
+        if entry.type == "favorite":
+            items.append(
+                {
+                    "type": "favorite",
+                    "timestamp": entry.timestamp,
+                }
+            )
+            continue
+
+        t = TrackStore.trackhashmap.get(entry.trackhash)
+
+        if t is None:
+            continue
+
+        item = {
+            "type": "track",
+            "hash": entry.trackhash,
+            "timestamp": entry.timestamp,
+        }
+        items.append(item)
+
+    return items
@@ -0,0 +1,42 @@
+from swingmusic.db.userdata import ScrobbleTable
+from swingmusic.lib.home.create_items import create_items
+from swingmusic.models.logger import TrackLog
+
+
+def get_recently_played(
+    limit: int, userid: int | None = None, _entries: list[TrackLog] = []
+):
+    """
+    Get the recently played items for the homepage.
+
+    Pass a list of track log entries to use a subset of the scrobble table.
+    """
+    # TODO: Paginate this
+    items = []
+
+    BATCH_SIZE = 200
+    current_index = 0
+
+    if len(_entries):
+        entries = _entries
+        limit = 1
+    else:
+        entries = ScrobbleTable.get_all(0, BATCH_SIZE, userid=userid)
+
+    max_iterations = 20
+    iterations = 0
+
+    while len(items) < limit and iterations < max_iterations:
+        items.extend(create_items(entries, limit))
+        current_index += BATCH_SIZE
+
+        if len(items) < limit:
+            entries = ScrobbleTable.get_all(
+                start=current_index + 1, limit=BATCH_SIZE, userid=userid
+            )
+            if not entries:
+                break
+
+        iterations += 1
+
+    return items
@@ -0,0 +1,217 @@
+import pathlib
+from datetime import datetime
+
+from swingmusic.lib.playlistlib import get_first_4_images
+from swingmusic.models.playlist import Playlist
+from swingmusic.models.track import Track
+
+from swingmusic.store.tracks import TrackStore
+from swingmusic.store.albums import AlbumStore
+from swingmusic.store.artists import ArtistStore
+
+from itertools import groupby
+
+from swingmusic.utils import flatten
+from swingmusic.utils.dates import (
+    create_new_date,
+    date_string_to_time_passed,
+)
+
+older_albums = set()
+older_artists = set()
+
+
+def calc_based_on_percent(items: list[str], total: int):
+    """
+    Checks if items is more than 85% of total items. Returns a boolean and the most common item.
+    """
+    most_common = max(items, key=items.count)
+    most_common_count = items.count(most_common)
+
+    return most_common_count / total >= 0.7, most_common, most_common_count
+
+
+def check_is_album_folder(tracks: list[Track]):
+    albumhashes = [t.albumhash for t in tracks]
+    return calc_based_on_percent(albumhashes, len(tracks))
+
+
+def check_is_artist_folder(tracks: list[Track]):
+    # INFO: flatten artist hashes using "-" as a separator
+    artisthashes = flatten([t.artisthashes for t in tracks])
+    return calc_based_on_percent(artisthashes, len(tracks))
+
+
+def check_is_track_folder(tracks: list[Track]):
+    # INFO: is more of a playlist
+    if len(tracks) >= 3:
+        return False
+
+    return [create_track(t) for t in tracks]
+
+
+def create_track(t: Track):
+    """
+    Creates a recently added track entry.
+    """
+    return {
+        "type": "track",
+        "hash": t.trackhash,
+        "timestamp": t.last_mod,
+        "help_text": "NEW TRACK",
+    }
+
+# INFO: Keys: folder, tracks, time (timestamp)
+# group_type = dict[str, str | list[Track] | float]
+
+
+def check_folder_type(group_: dict):
+    # check if all tracks in group have the same albumhash
+    # if so, return "album"
+    key: str = group_["folder"]
+    tracks: list[Track] = group_["tracks"]
+    time: float = group_["time"]
+    existing_artist_hashes: set[str] = set(ArtistStore.artistmap.keys())
+    existing_album_hashes: set[str] = set(AlbumStore.albummap.keys())
+
+    if len(tracks) == 1:
+        entry = create_track(tracks[0])
+        entry["timestamp"] = time
+        return entry
+
+    is_album, albumhash, _ = check_is_album_folder(tracks)
+    if is_album:
+        # album = AlbumTable.get_album_by_albumhash(albumhash)
+        entry = AlbumStore.albummap.get(albumhash)
+
+        if entry is None:
+            return None
+
+        return {
+            "type": "album",
+            "hash": albumhash,
+            "timestamp": time,
+            "help_text": (
+                "NEW ALBUM" if albumhash in existing_album_hashes else "NEW TRACKS"
+            ),
+        }
+
+    is_artist, artisthash, trackcount = check_is_artist_folder(tracks)
+    if is_artist:
+        entry = ArtistStore.artistmap.get(artisthash)
+
+        if entry is None:
+            return None
+
+        return {
+            "type": "artist",
+            "hash": artisthash,
+            "timestamp": time,
+            "help_text": (
+                "NEW ARTIST" if artisthash not in existing_artist_hashes else "NEW MUSIC"
+            ),
+        }
+
+    is_track_folder = check_is_track_folder(tracks)
+
+    return (
+        is_track_folder
+        if is_track_folder
+        else {
+            "type": "folder",
+            "hash": key,
+            "timestamp": time,
+            "help_text": "NEW MUSIC",
+        }
+    )
+
+
+def group_track_by_folders(tracks: list[Track], groups: dict[str, list[Track]]):
+    """
+    Groups tracks by folder and returns a list of groups sorted by last modified date.
+
+    Uses generator expressions to avoid creating intermediate lists.
+    """
+    # INFO: sort tracks by folder name, then group by folder name
+    tracks = sorted(tracks, key=lambda t: t.folder)
+    thisgroup = groupby(tracks, lambda t: t.folder)
+
+    for folder, thistracks in thisgroup:
+        groups.setdefault(folder, []).extend(thistracks)
+
+    return groups
+
+
+def get_recently_added_items(limit: int = 7):
+    tracks = get_recently_added_tracks(start=0, limit=None)
+    groups = group_track_by_folders(tracks, {})
+    grouplist = []
+
+    # INFO: sort tracks by last modified date in descending order to get the most recent last modified date
+    for folder, trackgroup in groups.items():
+        if not pathlib.Path(folder).exists():
+            continue
+
+        trackgroup.sort(key=lambda t: t.last_mod, reverse=True)
+        grouplist.append(
+            {
+                "folder": folder,
+                "len": len(trackgroup),
+                "tracks": trackgroup,
+                "time": trackgroup[0].last_mod,
+            }
+        )
+
+    # sort groups by last modified date
+    grouplist = sorted(grouplist, key=lambda group: group["time"], reverse=True)
+
+    recent_items = []
+
+    for group in grouplist:
+        item = check_folder_type(group)
+
+        if item not in recent_items:
+            if not item:
+                continue
+
+            (
+                recent_items.append(item)
+                if type(item) == dict
+                else recent_items.extend(item)
+            )
+
+        if len(recent_items) >= limit:
+            break
+
+    return recent_items
+
+
+def get_recently_added_playlist(limit: int = 100):
+    playlist = Playlist(
+        id="recentlyadded",
+        name="Recently Added",
+        image=None,
+        last_updated="Now",
+        settings={},
+        trackhashes=[],
+    )
+
+    tracks = get_recently_added_tracks(limit=limit)
+
+    try:
+        # Create date to show as last updated
+        date = datetime.fromtimestamp(tracks[0].last_mod)
+    except IndexError:
+        return playlist, []
+
+    playlist._last_updated = date_string_to_time_passed(create_new_date(date))
+    images = get_first_4_images(tracks=tracks)
+    playlist.images = images
+    playlist.duration = sum(t.duration for t in tracks)
+    playlist.count = len(tracks)
+
+    return playlist, tracks
+
+
+def get_recently_added_tracks(start: int = 0, limit: int | None = 100):
+    return TrackStore.get_recently_added(start, limit)
@@ -0,0 +1,35 @@
+from datetime import datetime
+
+from swingmusic.db.userdata import ScrobbleTable
+from swingmusic.models.playlist import Playlist
+from swingmusic.lib.playlistlib import get_first_4_images
+from swingmusic.utils.dates import (
+    create_new_date,
+    date_string_to_time_passed,
+)
+
+from swingmusic.store.tracks import TrackStore
+
+
+def get_recently_played_playlist(limit: int = 100):
+    playlist = Playlist(
+        id="recentlyplayed",
+        name="Recently Played",
+        image=None,
+        last_updated="Now",
+        settings={},
+        trackhashes=[],
+    )
+
+    scrobbles = ScrobbleTable.get_all(None, 100)
+    tracks = TrackStore.get_tracks_by_trackhashes(
+        [scrobble.trackhash for scrobble in scrobbles]
+    )
+
+    date = datetime.fromtimestamp(tracks[0].lastplayed)
+    playlist._last_updated = date_string_to_time_passed(create_new_date(date))
+
+    images = get_first_4_images(tracks=tracks)
+    playlist.images = images
+
+    return playlist, tracks
@@ -0,0 +1,161 @@
+from swingmusic.db.userdata import FavoritesTable, PlaylistTable
+from swingmusic.lib.home import find_mix
+from swingmusic.lib.home.recentlyadded import get_recently_added_playlist
+from swingmusic.lib.home.recentlyplayed import get_recently_played_playlist
+from swingmusic.lib.playlistlib import get_first_4_images
+from swingmusic.serializers.album import album_serializer
+from swingmusic.serializers.artist import serialize_for_card
+from swingmusic.serializers.playlist import serialize_for_card as serialize_playlist
+from swingmusic.serializers.track import serialize_track
+from swingmusic.store.albums import AlbumStore
+from swingmusic.store.artists import ArtistStore
+from swingmusic.store.folder import FolderStore
+from swingmusic.store.tracks import TrackStore
+from swingmusic.utils.dates import timestamp_to_time_passed
+
+
+def recover_items(items: list[dict]):
+    custom_playlists = [
+        {"name": "recentlyadded", "handler": get_recently_added_playlist},
+        {"name": "recentlyplayed", "handler": get_recently_played_playlist},
+    ]
+    recovered = []
+
+    for item in items:
+        recovered_item = None
+
+        if item["type"] == "album":
+            album = AlbumStore.get_album_by_hash(item["hash"])
+            if album is None:
+                continue
+
+            album = album_serializer(
+                album,
+                to_remove={
+                    "genres",
+                    "date",
+                    "count",
+                    "duration",
+                    "albumartists_hashes",
+                    "og_title",
+                },
+            )
+
+            recovered_item = {
+                "type": "album",
+                "item": album,
+            }
+        elif item["type"] == "artist":
+            artist = ArtistStore.get_artist_by_hash(item["hash"])
+            if artist is None:
+                continue
+
+            recovered_item = {
+                "type": "artist",
+                "item": serialize_for_card(artist),
+            }
+        elif item["type"] == "folder":
+            count = FolderStore.count_tracks_containing_paths([item["hash"]])
+
+            recovered_item = {
+                "type": "folder",
+                "item": {
+                    "path": item["hash"],
+                    "count": count[0]["trackcount"],
+                },
+            }
+        elif item["type"] == "playlist":
+            if item.get("is_custom"):
+                playlist, _ = next(
+                    i["handler"]()
+                    for i in custom_playlists
+                    if i["name"] == item["hash"]
+                )
+                playlist.images = [i["image"] for i in playlist.images]
+
+                playlist = serialize_playlist(
+                    playlist, to_remove={"settings", "duration"}
+                )
+                recovered_item = {
+                    "type": "playlist",
+                    "item": playlist,
+                }
+            else:
+                playlist = PlaylistTable.get_by_id(item["hash"])
+                if playlist is None:
+                    continue
+
+                tracks = TrackStore.get_tracks_by_trackhashes(playlist.trackhashes)
+                playlist.clear_lists()
+
+                if not playlist.has_image:
+                    images = get_first_4_images(tracks)
+                    images = [i["image"] for i in images]
+                    playlist.images = images
+
+                recovered_item = {
+                    "type": "playlist",
+                    "item": serialize_playlist(playlist),
+                }
+        elif item["type"] == "favorite":
+            image = None
+            last_trackhash = FavoritesTable.get_last_trackhash()
+
+            if last_trackhash:
+                trackhash = last_trackhash.replace("track_", "")
+                entry = TrackStore.trackhashmap.get(trackhash)
+                if entry:
+                    image = entry.tracks[0].image
+
+            recovered_item = {
+                "type": "favorite",
+                "item": {
+                    "count": FavoritesTable.count_tracks(),
+                    "image": image,
+                },
+            }
+        elif item["type"] == "track":
+            track = TrackStore.trackhashmap.get(item["hash"])
+            if track is None:
+                continue
+
+            recovered_item = {
+                "type": "track",
+                "item": serialize_track(track.get_best()),
+            }
+
+        elif item["type"] == "mix":
+            try:
+                splits = item["hash"].split(".")
+                mixid = splits[0]
+                sourcehash = splits[1]
+            except IndexError:
+                continue
+
+            mix = find_mix(mixid, sourcehash)
+            if mix is None:
+                continue
+
+            recovered_item = {
+                "type": "mix",
+                "item": mix,
+            }
+
+        if recovered_item is not None:
+            helptext = item.get("help_text") or item.get("type")
+            secondary_text = item.get("secondary_text")
+
+            if "secondary_text" in item:
+                secondary_text = item["secondary_text"]
+            elif "timestamp" in item:
+                secondary_text = timestamp_to_time_passed(item["timestamp"])
+
+            if helptext:
+                recovered_item["item"]["help_text"] = helptext
+
+            if secondary_text:
+                recovered_item["item"]["time"] = secondary_text
+
+            recovered.append(recovered_item)
+
+    return recovered
@@ -0,0 +1,43 @@
+import gc
+import logging
+from time import time
+from swingmusic.lib.mapstuff import (
+    map_album_colors,
+    map_artist_colors,
+    map_favorites,
+    map_scrobble_data,
+)
+from swingmusic.lib.populate import CordinateMedia
+from swingmusic.lib.recipes.recents import RecentlyAdded
+from swingmusic.lib.tagger import IndexTracks
+from swingmusic.store.albums import AlbumStore
+from swingmusic.store.artists import ArtistStore
+from swingmusic.store.folder import FolderStore
+from swingmusic.store.tracks import TrackStore
+from swingmusic.utils.threading import background
+
+log = logging.getLogger(__name__)
+
+@background
+def index_everything():
+    IndexTracks()
+
+    key = str(time())
+    TrackStore.load_all_tracks(key)
+    AlbumStore.load_albums(key)
+    ArtistStore.load_artists(key)
+    FolderStore.load_filepaths()
+
+    # NOTE: Rebuild recently added items on the homepage store
+    RecentlyAdded()
+
+    # map colors
+    map_album_colors()
+    map_artist_colors()
+
+    map_scrobble_data()
+    map_favorites()
+
+    CordinateMedia(instance_key=str(time()))
+    gc.collect()
+    log.info("Indexing completed")
@@ -0,0 +1,343 @@
+import datetime
+import pathlib
+from pathlib import Path
+
+from swingmusic.store.tracks import TrackStore
+
+
+# # # # # # # # # # # # # # # # # # # #
+# Functions for parsing lyrics lines  #
+# # # # # # # # # # # # # # # # # # # #
+
+def parse_lyrics_lines(lyrics:str) -> list[dict]:
+    """
+    Split lyrics into lines and determine there tag type.
+
+    Parses the tag if the following format is present: [tag]*[tags] <body>
+    else tag_type is unknown
+    tag-type and tags are lists combined by their index
+
+
+    :param lyrics: Full lyrics body
+    :return: {'tag_types', 'body', 'tags'}
+    """
+
+
+    entries = []
+    for line in lyrics.splitlines():
+
+        data = {
+            "tag_types": [],
+            "tags": []
+        }
+        if line.startswith("["):
+
+            # loop until all tags are parsed in line
+            while True:
+                if "[" in line and "]" in line: # second tag
+                    bracket_content, after_content = line.split("]", 1)
+                    bracket_content = bracket_content.removeprefix("[")
+
+                    data["tags"].append(bracket_content)
+                    data["body"] = after_content
+
+                    line = after_content
+
+                    # check which tag type it is
+                    if bracket_content[0].isnumeric():
+                        data["tag_types"].append( "time" )
+
+                    elif bracket_content[0].isalpha():
+                        data["tag_types"].append( "meta" )
+                else:
+                    # if no brackets inside the line, there is also no tag.
+                    break
+
+        elif line.startswith("#"):
+            data["tag_types"].append("comment")
+            data["tags"] = ""
+            data["body"] = line
+
+        else:
+            data["tag_types"].append("unknown")
+            data["tags"] = "unknown"
+            data["body"] = line
+
+        entries.append(data)
+
+    return entries
+
+
+def filter_parse_lyrics_lines(lines:list[dict], tag_types:list|str) -> list[dict]:
+    """
+    filter all lyrics line to only contain given tags
+
+    :param lines: list returned by `parse_lyrics_lines`
+    :param tag_types: list or string of tags return should contain
+    """
+
+    if isinstance(tag_types, str):
+        tag_types = [tag_types]
+
+    found_tags = []
+
+    # line = {"tags", "body", "tag_types"}
+    for line in lines:
+        group = {
+            "tag_types": [],
+            "tags": []
+        }
+        for (tag, tag_type) in zip(line["tags"], line["tag_types"]):
+            if tag_type in tag_types:
+                group["tag_types"].append(tag_type)
+                group["tags"].append(tag)
+                group["body"] = line["body"]
+
+        # filter out no match
+        if len(group["tags"]) > 0:
+            found_tags.append(group)
+
+    return found_tags
+
+
+def parse_time_tag(lines:list[dict]) -> list[dict]:
+    """
+    Filter time-tags from lines and parse them.
+    """
+
+    # filter tag-type time
+    # format into dict with timestamps
+
+    parsed_times = []
+    time_tags = filter_parse_lyrics_lines(lines, "time")
+
+    # line = {"tags", "body", "tag_types"}
+    for line in time_tags:
+        for (tag, tag_type) in zip(line["tags"], line["tag_types"]):
+            minute, seconds = tag.split(":", 1)
+
+            parsed_times.append({
+            "minute": minute,
+            "seconds": seconds,
+            "body": line["body"],
+            })
+
+    return parsed_times
+
+
+# # # # # # # # # # # # # # # # # # # #
+#  Lyrics class for simplified usage  #
+# # # # # # # # # # # # # # # # # # # #
+
+
+class Lyrics:
+
+    SUPPORTED_METATAGS = {
+        "ti": "title",
+        "ar": "artist",
+        "al": "album",
+        "au": "author",
+        "lr": "lyricist",
+        "length": "length",
+        "by": "lrc_author",
+        "offset": "offset",
+        "re": "recorder",
+        "tool": "tool",
+        "ve": "version"
+    }
+
+    lyrics:str
+    parsed_lyrics:list[dict]
+    meta:dict = {}
+
+    is_synced:bool = False
+
+
+    def __init__(self, lyrics:str=""):
+        """
+
+        :param lyrics: entire lyrics body
+        """
+
+        if lyrics is None:
+            raise ValueError("Lyrics can not be None")
+
+        if isinstance(lyrics, list):
+            lyrics = lyrics[0]
+
+        lyrics = lyrics.replace("engdesc", "")
+        self.lyrics = lyrics
+
+        parsed = parse_lyrics_lines(lyrics)
+
+        # translate meta tags
+        meta = filter_parse_lyrics_lines(parsed, "meta")
+        for line in meta:
+            for tag in line["tags"]:
+                name, body = tag.split(":", 1)
+                name = name.lower()
+
+                dict_name = self.SUPPORTED_METATAGS.get(name, name)
+                self.meta[dict_name] = body
+
+
+        # check if synced or not.
+        # not fail-save:
+        # If even just one time tag in the entire lyrics gets flagged as synced
+        if len(filter_parse_lyrics_lines(parsed, "time")) > 0:
+            self.is_synced = True
+            self.parsed_lyrics = filter_parse_lyrics_lines(parsed, "time")
+        else:
+            self.is_synced = False
+            self.parsed_lyrics = filter_parse_lyrics_lines(parsed, "unknown")
+
+        # TODO: add support for multilanguage lyrics
+
+
+    def format_synced_lyrics(self):
+        """
+        Formats synced lyrics into a list of dicts
+        """
+        if not self.is_synced:
+            raise ValueError("Cannot format synced lyrics if no synced lyrics exist for track.\nPlease use `format_unsynced_lyrics()`")
+
+        lyrics = []
+
+        time_tags = parse_time_tag(self.parsed_lyrics)
+
+        for entry in time_tags:
+            minutes = entry["minute"]
+            if "." in entry["seconds"]:
+                seconds = entry["seconds"].split(".")[0]
+                milli = entry["seconds"].split(".")[-1]
+            else:
+                seconds = entry["seconds"]
+                milli = "0"
+
+            minutes = int(minutes)
+            seconds = int(seconds)
+            milli = int(milli)
+
+            seconds = datetime.timedelta(minutes=minutes, seconds=seconds, milliseconds=milli).total_seconds()
+
+            offset = 0
+            if "offset" in self.meta:
+                offset = int(self.meta["offset"])  # offset in milliseconds
+
+            milliseconds = seconds * 1000 - offset
+            lyrics.append({"time": milliseconds, "text": entry["body"]})
+
+        return lyrics
+
+
+    def format_unsynced_lyrics(self) -> list[str]:
+        """
+        return unsynced lyrics.
+        If no lyrics provided return empty string.
+        """
+
+        lyrics = [item["body"] for item in self.parsed_lyrics]
+        return lyrics
+
+
+    def __bool__(self):
+        """
+        return True if contains anything
+        """
+        return bool(self.parsed_lyrics)
+
+
+# # # # # # # # # # # # # # # # # # # # # # # # # # #
+# Path and parse function to get lyrics from track  #
+# # # # # # # # # # # # # # # # # # # # # # # # # # #
+
+def get_lyrics_file(track_path: str|pathlib.Path) -> Lyrics:
+    """
+    Try to get lyrics from a relative lrc file.
+
+    :param track_path: path of track
+    """
+
+    track_path = Path(track_path)
+    lyrics_path = track_path.with_suffix(".lrc")
+    extended_path = track_path.with_suffix(".rlrc")
+
+    # check paths
+    if lyrics_path.exists():
+        lyrics = Lyrics(lyrics_path.read_text())
+        return lyrics
+
+    elif extended_path.exists():
+        lyrics = Lyrics(extended_path.read_text())
+        return lyrics
+
+    else:
+        return Lyrics()
+
+
+def get_lyrics_from_duplicates(track_path: str, trackhash: str) -> Lyrics:
+    """
+    Finds the lyrics from other duplicate tracks
+
+    :param track_path: path of track
+    :param trackhash: Track-hash value
+    """
+
+    entry = TrackStore.trackhashmap.get(trackhash, None)
+
+    if entry is None:
+        return Lyrics()
+
+    for track in entry.tracks:
+        if track.trackhash == trackhash and track.filepath != track_path:
+            lyrics = get_lyrics_file(track.filepath)
+
+            if lyrics:
+                return lyrics
+
+    return Lyrics()
+
+
+def get_lyrics_from_tags(trackhash: str) -> Lyrics:
+    """
+    Gets the lyrics from the tags of the track
+
+    :param trackhash:
+    """
+
+    entry = TrackStore.trackhashmap.get(trackhash, None)
+
+    if entry is None:
+        return Lyrics()
+
+    for track in entry.tracks:
+        if "lyrics" in track.extra:
+            lyrics = track.extra["lyrics"]
+            if lyrics:
+                return Lyrics(lyrics)
+
+    return Lyrics("")
+
+
+def check_lyrics_file(filepath: str, trackhash: str):
+    """
+    Checks if the lyrics file exists for a track
+    """
+
+    lyrics_file = Path(filepath).with_suffix(".lrc")
+    if lyrics_file.exists:
+        return True
+
+    entry = TrackStore.trackhashmap.get(trackhash, None)
+
+    if entry is None:
+        return False
+
+    for track in entry.tracks:
+        if track.trackhash == trackhash and track.filepath != filepath:
+            lyrics_file = Path(track.filepath).with_suffix(".lrc")
+
+            if lyrics_file.exists():
+                return True
+
+    return False
@@ -0,0 +1,94 @@
+from swingmusic.db.userdata import LibDataTable, FavoritesTable, ScrobbleTable
+from swingmusic.store.albums import AlbumStore
+from swingmusic.store.artists import ArtistStore
+from swingmusic.store.tracks import TrackStore
+
+
+from typing import Any
+
+
+def map_scrobble_data():
+    """
+    Maps scrobble data to the in-memory stores.
+
+    The scrobble data is loaded from the database and grouped by trackhash.
+    The album and artist scrobble data (for those tracks) are then incremented based on the data.
+    """
+    records = ScrobbleTable.get_all(0, None)
+
+    # group records by trackhash
+    grouped: dict[str, dict[str, Any]] = {}
+
+    for record in records:
+        # aggregate playcount, playduration and lastplayed
+        item = grouped.setdefault(record.trackhash, {})
+        item["playcount"] = item.get("playcount", 0) + 1
+        item["playduration"] = item.get("playduration", 0) + record.duration
+        item["lastplayed"] = max(item.get("lastplayed", 0), record.timestamp)
+
+    # increment playcount, playduration and lastplayed for albums and artists
+    for trackhash, data in grouped.items():
+        track = TrackStore.trackhashmap.get(trackhash)
+
+        if track is None:
+            continue
+
+        track.increment_playcount(
+            data["playduration"], data["lastplayed"], data["playcount"]
+        )
+
+        album = AlbumStore.albummap.get(track.tracks[0].albumhash)
+        if album:
+            album.increment_playcount(
+                data["playduration"], data["lastplayed"], data["playcount"]
+            )
+
+        for artisthash in track.tracks[0].artisthashes:
+            artist = ArtistStore.artistmap.get(artisthash)
+            if artist:
+                artist.increment_playcount(
+                    data["playduration"], data["lastplayed"], data["playcount"]
+                )
+
+
+def map_favorites():
+    """
+    Maps favorites data to the in-memory stores.
+    """
+    favorites = FavoritesTable.get_all()
+
+    for entry in favorites:
+        if entry.type == "album":
+            album = AlbumStore.albummap.get(entry.hash)
+            if album:
+                album.toggle_favorite_user(entry.userid)
+
+        elif entry.type == "artist":
+            artist = ArtistStore.artistmap.get(entry.hash)
+            if artist:
+                artist.toggle_favorite_user(entry.userid)
+
+        elif entry.type == "track":
+            track = TrackStore.trackhashmap.get(entry.hash)
+            if track:
+                track.toggle_favorite_user(entry.userid)
+
+
+def map_artist_colors():
+    colors = LibDataTable.get_all_colors(type="artist")
+
+    for color in colors:
+        artist = ArtistStore.artistmap.get(color["itemhash"])
+
+        if artist:
+            artist.set_color(color["color"])
+
+
+def map_album_colors():
+    colors = LibDataTable.get_all_colors(type="album")
+
+    for color in colors:
+        album = AlbumStore.albummap.get(color["itemhash"])
+
+        if album:
+            album.set_color(color["color"])
@@ -0,0 +1,76 @@
+import json
+from typing import Any
+from swingmusic.serializers.album import serialize_for_card
+from swingmusic.serializers.artist import serialize_for_card as serialize_artist
+from swingmusic.store.albums import AlbumStore
+from swingmusic.store.artists import ArtistStore
+from swingmusic.utils.hashing import create_hash
+
+
+def validate_page_items(items: list[dict[str, str]], existing: list[dict[str, str]]):
+    """
+    Validate the items in a page before adding them to the database.
+    """
+    validated: list[dict[str, str]] = []
+    indexed = set(create_hash(json.dumps(item)) for item in existing)
+
+    for item in items:
+        if create_hash(json.dumps(item)) in indexed:
+            continue
+
+        if item["type"] == "album":
+            album = AlbumStore.albummap.get(item["hash"])
+
+            if album is not None:
+                validated.append(item)
+        elif item["type"] == "artist":
+            artist = ArtistStore.artistmap.get(item["hash"])
+
+            if artist is not None:
+                validated.append(item)
+        else:
+            raise ValueError(f"Invalid item type: {item['type']}")
+
+    return validated
+
+
+def remove_page_items(existing: list[dict[str, str]], item: dict[str, str]):
+    return [
+        i
+        for i in existing
+        if create_hash(json.dumps(i)) != create_hash(json.dumps(item))
+    ]
+
+
+def recover_page_items(items: list[dict[str, str]], for_homepage: bool = False):
+    """
+    Recover the items in a page.
+    """
+    recovered: list[dict[str, Any]] = []
+
+    for item in items:
+        if item["type"] == "album":
+            album = AlbumStore.albummap.get(item["hash"])
+
+            if album is not None:
+                item = serialize_for_card(album.album)
+
+                if for_homepage:
+                    del item["type"]
+                    item = {"item": item, "type": "album"}
+
+                recovered.append(item)
+        elif item["type"] == "artist":
+            artist = ArtistStore.artistmap.get(item["hash"])
+
+            if artist is not None:
+                item = serialize_artist(artist.artist)
+
+                if for_homepage:
+                    del item["type"]
+                    item = {"item": item, "type": "artist"}
+
+                recovered.append(item)
+
+    recovered.reverse()
+    return recovered
@@ -0,0 +1,171 @@
+"""
+This library contains all the functions related to playlists.
+"""
+import random
+import string
+import logging
+from PIL import Image, ImageSequence
+
+from swingmusic import settings
+from swingmusic.models.track import Track
+from swingmusic.store.albums import AlbumStore
+from swingmusic.store.tracks import TrackStore
+
+logger = logging.getLogger(__name__)
+
+def create_thumbnail(image: Image, img_name: str) -> str:
+    """
+    Creates a 250 px high thumbnail from the Image.
+    It will keep the aspect ratio.
+
+    Images are saved in the playlist-img path
+
+    :param image: Image object.
+    :param img_name: Name of image.
+    :return: Filename of image.
+    """
+
+    aspect_ratio = image.width / image.height
+    new_w = round(250 * aspect_ratio)
+    thumb = image.resize((new_w, 250), Image.Resampling.LANCZOS)
+
+    thumb_filename = "thumb_" + img_name
+    thumb_path = settings.Paths().playlist_img_path / thumb_filename
+
+    thumb.save(thumb_path, "webp")
+
+    return thumb_filename
+
+
+def create_gif_thumbnail(image: Image, img_name: str):
+    """
+    Creates a 250 px high thumbnail from the provided GIF.
+    Keeps the aspect ratio.
+
+    Images are saved in the playlist-img path
+
+    :param image: Image object.
+    :param img_name: Name of image.
+    :return: Filename of image.
+    """
+    thumb_name = "thumb_" + img_name
+    thumb_path = settings.Paths().playlist_img_path / thumb_name
+
+    frames = []
+    for frame in ImageSequence.Iterator(image):
+        aspect_ratio = frame.width / frame.height
+        new_w = round(250 * aspect_ratio)
+        thumb = frame.resize((new_w, 250), Image.Resampling.LANCZOS)
+
+        frames.append(thumb)
+
+    frames[0].save(thumb_path, save_all=True, append_images=frames[1:])
+
+    return thumb_name
+
+
+def save_p_image(img: Image, pid: int, content_type: str = None, filename: str = None) -> str:
+    """
+    Saves a playlist banner image and returns the filepath.
+    """
+    # img = Image.open(file)
+
+    random_str = "".join(random.choices(string.ascii_letters + string.digits, k=5))
+
+    if not filename:
+        filename = str(pid) + str(random_str) + ".webp"
+
+    full_img_path = settings.Paths().playlist_img_path / filename
+
+    if content_type == "image/gif":
+        frames = []
+
+        for frame in ImageSequence.Iterator(img):
+            frames.append(frame.copy())
+
+        frames[0].save(full_img_path, save_all=True, append_images=frames[1:])
+        create_gif_thumbnail(img, img_path=filename)
+
+        return filename
+
+    img.save(full_img_path, "webp")
+    create_thumbnail(img, img_name=filename)
+
+    return filename
+
+
+def duplicate_images(images: list):
+    if len(images) == 1:
+        images *= 4
+    elif len(images) == 2:
+        images += list(reversed(images))
+    elif len(images) == 3:
+        images = images + images[:1]
+
+    return images
+
+# TODO: mutable var in param.
+def get_first_4_images(
+        tracks: list[Track] = [],
+        trackhashes: list[str] = []
+) -> list[dict["str", str]]:
+    """
+    Returns images of the first 4 albums that appear in the track list.
+
+    When tracks are not passed, trackhashes need to be passed.
+    Tracks are then resolved from the store.
+    """
+    if len(trackhashes) > 0:
+        tracks = TrackStore.get_tracks_by_trackhashes(trackhashes)
+
+    albums = []
+
+    for track in tracks:
+        if track.albumhash not in albums:
+            albums.append(track.albumhash)
+
+            if len(albums) == 4:
+                break
+
+    albums = AlbumStore.get_albums_by_hashes(albums)
+    images = [
+        {
+            "image": album.image,
+            "color": album.color,
+        }
+        for album in albums
+    ]
+
+    if len(images) == 4:
+        return images
+
+    return duplicate_images(images)
+
+
+def cleanup_playlist_images() -> None:
+    """
+    Deletes all unlinked files in playlist-img folder.
+    All files not present in the PlaylistTable will get deleted
+    """
+    # Import here to avoid circular import
+    from swingmusic.db.userdata import PlaylistTable
+
+    playlists = PlaylistTable.get_all()
+    linked_images = {p.image for p in playlists if p.image and p.image != "None"}
+
+    playlist_dir = settings.Paths().playlist_img_path
+
+    # Find unlinked images (including thumbnails)
+    for file in playlist_dir.iterdir():
+        if not file.isfile:
+            continue
+
+        name = file.name # not stem. PlaylistTable saves with extension
+        if file not in linked_images:
+            if name.removeprefix("thumb_") not in linked_images:
+                continue
+
+            try:
+                file.unlink(missing_ok=True)
+            except OSError as e:
+                logger.exception("could not delete file", exc_info=e)
@@ -0,0 +1,176 @@
+import functools
+import os
+from dataclasses import asdict
+import multiprocessing as mp
+from requests import ReadTimeout
+from concurrent.futures import ProcessPoolExecutor
+from requests import ConnectionError as RequestConnectionError
+import logging
+
+from swingmusic import settings
+from swingmusic.lib.artistlib import CheckArtistImages
+from swingmusic.lib.taglib import extract_thumb
+from swingmusic.models import Album, Artist
+from swingmusic.models.lastfm import SimilarArtist
+from swingmusic.models.track import Track
+from swingmusic.store.albums import AlbumStore
+from swingmusic.store.artists import ArtistStore
+from swingmusic.utils.network import has_connection
+from swingmusic.utils.progressbar import tqdm
+from swingmusic.request.artists import fetch_similar_artists
+from swingmusic.lib.colorlib import ProcessAlbumColors, ProcessArtistColors
+
+from swingmusic.db.userdata import SimilarArtistTable
+
+log = logging.getLogger(__name__)
+
+
+class CordinateMedia:
+    """
+    Cordinates the extracting of thumbnails
+    """
+
+    def __init__(self, instance_key: str):
+        ProcessTrackThumbnails()
+        ProcessAlbumColors()
+        ProcessArtistColors()
+
+        tried_to_download_new_images = False
+
+        if has_connection():
+            tried_to_download_new_images = True
+            try:
+                CheckArtistImages()
+            except (RequestConnectionError, ReadTimeout) as e:
+                log.error(
+                    "Internet connection lost. Downloading artist images suspended."
+                )
+                log.error(e)  # REVIEW More informations = good
+        else:
+            log.warning("No internet connection. Downloading artist images suspended!")
+
+        # Re-process the new artist images.
+        if tried_to_download_new_images:
+            ProcessArtistColors()
+
+        if has_connection():
+            print("Attempting to download similar artists...")
+            FetchSimilarArtistsLastFM()
+
+
+def get_image(tracks: list[Track], paths=None):
+    """
+    The function retrieves an image from a list of tracks by extracting the thumbnail from the first track that has one.
+
+    :param tracks: A list of Track objects to extract the image from.
+    :type tracks: list[Track]
+    :return: None
+    """
+
+    for track in tracks:
+        extracted = extract_thumb(track.filepath, track.albumhash + ".webp", paths)
+
+        if extracted:
+            return
+
+
+class ProcessTrackThumbnails:
+    """
+    Extracts the album art from all albums in album store.
+    """
+
+    def extract(self, albums: list[Album]):
+        """
+        Extracts the album art with platform-specific logic.
+        """
+
+        cpus = max(1, os.cpu_count() // 2)
+
+        albumsMap = ( AlbumStore.get_album_tracks(album.albumhash) for album in albums )
+
+        # Create process pool with worker function
+        with mp.Pool(processes=cpus) as pool:
+            worker = functools.partial(get_image, paths=settings.Paths())
+            # Process files and track progress
+
+            results = list(
+                tqdm(
+                    pool.imap_unordered(worker, albumsMap),
+                    total=len(albums),
+                    desc="Extracting track images",
+                )
+            )
+
+            list(results)
+
+    def __init__(self) -> None:
+        """
+        Filters out albums that already have thumbnails and
+        extracts the thumbnail for the other albums.
+        """
+        path = settings.Paths().sm_thumb_path
+
+        # read all the files in the thumbnail directory
+        processed = set(file.stem for file in path.iterdir())
+        # filter out albums that already have thumbnails
+        albums = filter(
+            lambda album: album.albumhash not in processed,
+            AlbumStore.get_flat_list(),
+        )
+
+        albums = list(albums)
+        self.extract(albums)
+
+
+def save_similar_artists(artist: Artist):
+    """
+    Downloads and saves similar artists to the database.
+    """
+    if SimilarArtistTable.exists(artist.artisthash):
+        return
+
+    artists = fetch_similar_artists(artist.name)
+
+    # INFO: Nones mean there was a connection error
+    if artists is None:
+        return
+
+    artist_ = SimilarArtist(artist.artisthash, artists)
+    SimilarArtistTable.insert_one(asdict(artist_))
+
+
+class FetchSimilarArtistsLastFM:
+    """
+    Fetches similar artists from LastFM using a thread pool.
+    """
+
+    def __init__(self) -> None:
+        # read all artists from db
+        storeArtists = ArtistStore.get_flat_list()
+        processed = set(a.artisthash for a in SimilarArtistTable.get_all())
+
+        # filter out artists that already have similar artists using generator
+        def artist_generator():
+            for artist in storeArtists:
+                if artist.artisthash in processed:
+                    yield artist
+
+        artists = list(artist_generator())
+        cpus = max(1, os.cpu_count() // 2)
+
+        with ProcessPoolExecutor(max_workers=cpus) as executor:
+            try:
+                # TODO: fix negative total length
+                results = list(
+                    tqdm(
+                        executor.map(save_similar_artists, artist_generator()),
+                        total=len(artists),
+                        desc="Fetching similar artists",
+                    )
+                )
+
+                list(results)
+            # any exception that can be raised by the pool
+            except Exception as e:
+                log.warning(e)
+                return
@@ -0,0 +1,17 @@
+### Steps to reproduce
+
+### Expected behavior
+Tell us what should happen
+
+### Actual behavior
+Tell us what happens instead
+
+### Your System configuration
+- Python version: 
+- Pydub version: 
+- ffmpeg or avlib?: 
+- ffmpeg/avlib version: 
+
+### Is there an audio file you can include to help us reproduce?
+You can include the audio file in this issue - just put it in a zip file and drag/drop the zip file into the github issue.
+
@@ -0,0 +1,19 @@
+os: linux
+dist: bionic  # focal
+language: python
+before_install:
+  - sudo apt-get update --fix-missing
+install:
+  - sudo apt-get install -y ffmpeg libopus-dev python-scipy python3-scipy
+python:
+  - "2.7"
+  - "3.6"
+  - "3.7"
+  - "3.8"
+  - "3.9"
+  - "pypy2"
+  - "pypy3"
+script:
+  - python test/test.py
+after_script:
+  - pip install pylama && python -m pylama -i W,E501 pydub/ || true
@@ -0,0 +1,693 @@
+# API Documentation
+
+This document is a work in progress.
+
+If you're looking for some functionality in particular, it's a good idea to take a look at the [source code](https://github.com/jiaaro/pydub). Core functionality is mostly in `pydub/audio_segment.py` – a number of `AudioSegment` methods are in the `pydub/effects.py` module, and added to `AudioSegment` via the effect registration process (the `register_pydub_effect()` decorator function)
+
+Currently Undocumented:
+
+- Playback (`pydub.playback`)
+- Signal Processing (compression, EQ, normalize, speed change - `pydub.effects`, `pydub.scipy_effects`)
+- Signal generators (Sine, Square, Sawtooth, Whitenoise, etc - `pydub.generators`)
+- Effect registration system (basically the `pydub.utils.register_pydub_effect` decorator)
+
+
+## AudioSegment()
+
+`AudioSegment` objects are immutable, and support a number of operators.
+
+```python
+from pydub import AudioSegment
+sound1 = AudioSegment.from_file("/path/to/sound.wav", format="wav")
+sound2 = AudioSegment.from_file("/path/to/another_sound.wav", format="wav")
+
+# sound1 6 dB louder, then 3.5 dB quieter
+louder = sound1 + 6
+quieter = sound1 - 3.5
+
+# sound1, with sound2 appended
+combined = sound1 + sound2
+
+# sound1 repeated 3 times
+repeated = sound1 * 3
+
+# duration
+duration_in_milliseconds = len(sound1)
+
+# first 5 seconds of sound1
+beginning = sound1[:5000]
+
+# last 5 seconds of sound1
+end = sound1[-5000:]
+
+# split sound1 in 5-second slices
+slices = sound1[::5000]
+
+# Advanced usage, if you have raw audio data:
+sound = AudioSegment(
+    # raw audio data (bytes)
+    data=b'…',
+
+    # 2 byte (16 bit) samples
+    sample_width=2,
+
+    # 44.1 kHz frame rate
+    frame_rate=44100,
+
+    # stereo
+    channels=2
+)
+```
+
+Any operations that combine multiple `AudioSegment` objects in *any* way will first ensure that they have the same number of channels, frame rate, sample rate, bit depth, etc. When these things do not match, the lower quality sound is modified to match the quality of the higher quality sound so that quality is not lost: mono is converted to stereo, bit depth and frame rate/sample rate are increased as needed. If you do not want this behavior, you may explicitly reduce the number of channels, bits, etc using the appropriate `AudioSegment` methods.
+
+### AudioSegment(…).from_file()
+
+Open an audio file as an `AudioSegment` instance and return it. there are also a number of wrappers provided for convenience, but you should probably just use this directly.
+
+```python
+from pydub import AudioSegment
+
+# wave and raw don’t use ffmpeg
+wav_audio = AudioSegment.from_file("/path/to/sound.wav", format="wav")
+raw_audio = AudioSegment.from_file("/path/to/sound.raw", format="raw",
+                                   frame_rate=44100, channels=2, sample_width=2)
+
+# all other formats use ffmpeg
+mp3_audio = AudioSegment.from_file("/path/to/sound.mp3", format="mp3")
+
+# use a file you've already opened (advanced …ish)
+with open("/path/to/sound.wav", "rb") as wav_file:
+    audio_segment = AudioSegment.from_file(wav_file, format="wav")
+
+# also supports the os.PathLike protocol for python >= 3.6
+from pathlib import Path
+wav_path = Path("path/to/sound.wav")
+wav_audio = AudioSegment.from_file(wav_path)
+```
+
+The first argument is the path (as a string) of the file to read, **or** a file handle to read from.
+
+**Supported keyword arguments**:
+
+- `format` | example: `"aif"` | default: autodetected
+  Format of the output file. Supports `"wav"` and `"raw"` natively, requires ffmpeg for all other formats. `"raw"` files require 3 additional keyword arguments, `sample_width`, `frame_rate`, and `channels`, denoted below with: **`raw` only**. This extra info is required because raw audio files do not have headers to include this info in the file itself like wav files do.
+- `sample_width` | example: `2`
+  **`raw` only** — Use `1` for 8-bit audio `2` for 16-bit (CD quality) and `4` for 32-bit. It’s the number of bytes per sample.
+- `channels` | example: `1`
+  **`raw` only** — `1` for mono, `2` for stereo.
+- `frame_rate` | example: `2`
+  **`raw` only** — Also known as sample rate, common values are `44100` (44.1kHz - CD audio), and `48000` (48kHz - DVD audio)
+- `start_second` | example: `2.0` | default: `None`
+  Offset (in seconds) to start loading the audio file. If `None`, the audio will start loading from the beginning.
+- `duration` | example: `2.5` | default: `None`
+  Number of seconds to be loaded. If `None`, full audio will be loaded.
+
+
+### AudioSegment(…).export()
+
+Write the `AudioSegment` object to a file – returns a file handle of the output file (you don't have to do anything with it, though).
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file("/path/to/sound.wav", format="wav")
+
+# simple export
+file_handle = sound.export("/path/to/output.mp3", format="mp3")
+
+# more complex export
+file_handle = sound.export("/path/to/output.mp3",
+                           format="mp3",
+                           bitrate="192k",
+                           tags={"album": "The Bends", "artist": "Radiohead"},
+                           cover="/path/to/albumcovers/radioheadthebends.jpg")
+
+# split sound in 5-second slices and export
+for i, chunk in enumerate(sound[::5000]):
+  with open("sound-%s.mp3" % i, "wb") as f:
+    chunk.export(f, format="mp3")
+```
+
+The first argument is the location (as a string) to write the output, **or** a file handle to write to. If you do not pass an output file or path, a temporary file is generated.
+
+**Supported keyword arguments**:
+
+- `format` | example: `"aif"` | default: `"mp3"`
+  Format of the output file. Supports `"wav"` and `"raw"` natively, requires ffmpeg for all other formats.
+- `codec` | example: `"libvorbis"`
+  For formats that may contain content encoded with different codecs, you can specify the codec you'd like the encoder to use. For example, the "ogg" format is often used with the "libvorbis" codec. (requires ffmpeg)
+- `bitrate` | example: `"128k"`
+  For compressed formats, you can pass the bitrate you'd like the encoder to use (requires ffmpeg). Each codec accepts different bitrate arguments so take a look at the [ffmpeg documentation](https://www.ffmpeg.org/ffmpeg-codecs.html#Audio-Encoders) for details (bitrate usually shown as `-b`, `-ba` or `-a:b`).
+- `tags` | example: `{"album": "1989", "artist": "Taylor Swift"}`
+  Allows you to supply media info tags for the encoder (requires ffmpeg). Not all formats can receive tags (mp3 can).
+- `parameters` | example: `["-ac", "2"]`
+  Pass additional [command line parameters](https://www.ffmpeg.org/ffmpeg.html) to the ffmpeg call. These are added to the end of the call (in the output file section).
+- `id3v2_version` | example: `"3"` | default: `"4"`
+  Set the ID3v2 version used by ffmpeg to add tags to the output file. If you want Windows Exlorer to display tags, use `"3"` here ([source](http://superuser.com/a/453133)).
+- `cover` | example: `"/path/to/imgfile.png"`
+  Allows you to supply a cover image (path to the image file). Currently, only MP3 files allow this keyword argument. Cover image must be a jpeg, png, bmp, or tiff file.
+
+
+### AudioSegment.empty()
+
+Creates a zero-duration `AudioSegment`.
+
+```python
+from pydub import AudioSegment
+empty = AudioSegment.empty()
+
+len(empty) == 0
+```
+
+This is useful for aggregation loops:
+```python
+from pydub import AudioSegment
+
+sounds = [
+  AudioSegment.from_wav("sound1.wav"),
+  AudioSegment.from_wav("sound2.wav"),
+  AudioSegment.from_wav("sound3.wav"),
+]
+
+playlist = AudioSegment.empty()
+for sound in sounds:
+  playlist += sound
+```
+
+### AudioSegment.silent()
+
+Creates a silent audiosegment, which can be used as a placeholder, spacer, or as a canvas to overlay other sounds on top of.
+
+```python
+from pydub import AudioSegment
+
+ten_second_silence = AudioSegment.silent(duration=10000)
+```
+
+**Supported keyword arguments**:
+
+- `duration` | example: `3000` | default: `1000` (1 second)
+  Length of the silent `AudioSegment`, in milliseconds
+- `frame_rate` | example `44100` | default: `11025` (11.025 kHz)
+  Frame rate (i.e., sample rate) of the silent `AudioSegment` in Hz
+
+### AudioSegment.from_mono_audiosegments()
+
+Creates a multi-channel audiosegment out of multiple mono audiosegments (two or more). Each mono audiosegment passed in should be exactly the same length, down to the frame count.
+
+```python
+from pydub import AudioSegment
+
+left_channel = AudioSegment.from_wav("sound1.wav")
+right_channel = AudioSegment.from_wav("sound1.wav")
+
+stereo_sound = AudioSegment.from_mono_audiosegments(left_channel, right_channel)
+```
+
+### AudioSegment(…).dBFS
+
+Returns the loudness of the `AudioSegment` in dBFS (db relative to the maximum possible loudness). A Square wave at maximum amplitude will be roughly 0 dBFS (maximum loudness), whereas a Sine Wave at maximum amplitude will be roughly -3 dBFS.
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file("sound1.wav")
+
+loudness = sound.dBFS
+```
+
+### AudioSegment(…).channels
+
+Number of channels in this audio segment (1 means mono, 2 means stereo)
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file("sound1.wav")
+
+channel_count = sound.channels
+```
+
+### AudioSegment(…).sample_width
+
+Number of bytes in each sample (1 means 8 bit, 2 means 16 bit, etc). CD Audio is 16 bit, (sample width of 2 bytes).
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file("sound1.wav")
+
+bytes_per_sample = sound.sample_width
+```
+
+### AudioSegment(…).frame_rate
+
+CD Audio has a 44.1kHz sample rate, which means `frame_rate` will be `44100` (same as sample rate, see `frame_width`). Common values are `44100` (CD), `48000` (DVD), `22050`, `24000`, `12000` and `11025`.
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file("sound1.wav")
+
+frames_per_second = sound.frame_rate
+```
+
+### AudioSegment(…).frame_width
+
+Number of bytes for each "frame". A frame contains a sample for each channel (so for stereo you have 2 samples per frame, which are played simultaneously). `frame_width` is equal to `channels * sample_width`. For CD Audio it'll be `4` (2 channels times 2 bytes per sample).
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file("sound1.wav")
+
+bytes_per_frame = sound.frame_width
+```
+
+### AudioSegment(…).rms
+
+A measure of loudness. Used to compute dBFS, which is what you should use in most cases. Loudness is logarithmic (rms is not), which makes dB a much more natural scale.
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file("sound1.wav")
+
+loudness = sound.rms
+```
+
+### AudioSegment(…).max
+
+The highest amplitude of any sample in the `AudioSegment`. Useful for things like normalization (which is provided in `pydub.effects.normalize`).
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file("sound1.wav")
+
+peak_amplitude = sound.max
+```
+
+### AudioSegment(…).max_dBFS
+
+The highest amplitude of any sample in the `AudioSegment`, in dBFS (relative to the highest possible amplitude value). Useful for things like normalization (which is provided in `pydub.effects.normalize`).
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file("sound1.wav")
+
+normalized_sound = sound.apply_gain(-sound.max_dBFS)
+```
+
+### AudioSegment(…).duration_seconds
+
+Returns the duration of the `AudioSegment` in seconds (`len(sound)` returns milliseconds). This is provided for convenience; it calls `len()` internally.
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file("sound1.wav")
+
+assert sound.duration_seconds == (len(sound) / 1000.0)
+```
+
+### AudioSegment(…).raw_data
+
+The raw audio data of the AudioSegment. Useful for interacting with other audio libraries or weird APIs that want audio data in the form of a bytestring. Also comes in handy if you’re implementing effects or other direct signal processing.
+
+You probably don’t need this, but if you do… you’ll know.
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file("sound1.wav")
+
+raw_audio_data = sound.raw_data
+```
+
+### AudioSegment(…).frame_count()
+
+Returns the number of frames in the `AudioSegment`. Optionally you may pass in a `ms` keywork argument to retrieve the number of frames in that number of milliseconds of audio in the `AudioSegment` (useful for slicing, etc).
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file("sound1.wav")
+
+number_of_frames_in_sound = sound.frame_count()
+
+number_of_frames_in_200ms_of_sound = sound.frame_count(ms=200)
+```
+
+**Supported keyword arguments**:
+
+- `ms` | example: `3000` | default: `None` (entire duration of `AudioSegment`)
+  When specified, method returns number of frames in X milliseconds of the `AudioSegment`
+
+### AudioSegment(…).append()
+
+Returns a new `AudioSegment`, created by appending another `AudioSegment` to this one (i.e., adding it to the end), Optionally using a crossfade. `AudioSegment(…).append()` is used internally when adding `AudioSegment` objects together with the `+` operator.
+
+By default a 100ms (0.1 second) crossfade is used to eliminate pops and crackles.
+
+```python
+from pydub import AudioSegment
+sound1 = AudioSegment.from_file("sound1.wav")
+sound2 = AudioSegment.from_file("sound2.wav")
+
+# default 100 ms crossfade
+combined = sound1.append(sound2)
+
+# 5000 ms crossfade
+combined_with_5_sec_crossfade = sound1.append(sound2, crossfade=5000)
+
+# no crossfade
+no_crossfade1 = sound1.append(sound2, crossfade=0)
+
+# no crossfade
+no_crossfade2 = sound1 + sound2
+```
+
+**Supported keyword arguments**:
+
+- `crossfade` | example: `3000` | default: `100` (entire duration of `AudioSegment`)
+  When specified, method returns number of frames in X milliseconds of the `AudioSegment`
+
+### AudioSegment(…).overlay()
+
+Overlays an `AudioSegment` onto this one. In the resulting `AudioSegment` they will play simultaneously. If the overlaid `AudioSegment` is longer than this one, the result will be truncated (so the end of the overlaid sound will be cut off). The result is always the same length as this `AudioSegment` even when using the `loop`, and `times` keyword arguments.
+
+Since `AudioSegment` objects are immutable, you can get around this by overlaying the shorter sound on the longer one, or by creating a silent `AudioSegment` with the appropriate duration, and overlaying both sounds on to that one.
+
+```python
+from pydub import AudioSegment
+sound1 = AudioSegment.from_file("sound1.wav")
+sound2 = AudioSegment.from_file("sound2.wav")
+
+played_togther = sound1.overlay(sound2)
+
+sound2_starts_after_delay = sound1.overlay(sound2, position=5000)
+
+volume_of_sound1_reduced_during_overlay = sound1.overlay(sound2, gain_during_overlay=-8)
+
+sound2_repeats_until_sound1_ends = sound1.overlay(sound2, loop=true)
+
+sound2_plays_twice = sound1.overlay(sound2, times=2)
+
+# assume sound1 is 30 sec long and sound2 is 5 sec long:
+sound2_plays_a_lot = sound1.overlay(sound2, times=10000)
+len(sound1) == len(sound2_plays_a_lot)
+```
+
+**Supported keyword arguments**:
+
+- `position` | example: `3000` | default: `0` (beginning of this `AudioSegment`)
+  The overlaid `AudioSegment` will not begin until X milliseconds have passed
+- `loop` | example: `True` | default: `False` (entire duration of `AudioSegment`)
+  The overlaid `AudioSegment` will repeat (starting at `position`) until the end of this `AudioSegment`
+- `times` | example: `4` | default: `1` (entire duration of `AudioSegment`)
+  The overlaid `AudioSegment` will repeat X times (starting at `position`) but will still be truncated to the length of this `AudioSegment`
+- `gain_during_overlay` | example: `-6.0` | default: `0` (no change in volume during overlay)
+  Change the original audio by this many dB while overlaying audio. This can be used to make the original audio quieter while the overlaid audio plays.
+
+### AudioSegment(…).apply_gain(`gain`)
+
+Change the amplitude (generally, loudness) of the `AudioSegment`. Gain is specified in dB. This method is used internally by the `+` operator.
+
+```python
+from pydub import AudioSegment
+sound1 = AudioSegment.from_file("sound1.wav")
+
+# make sound1 louder by 3.5 dB
+louder_via_method = sound1.apply_gain(+3.5)
+louder_via_operator = sound1 + 3.5
+
+# make sound1 quieter by 5.7 dB
+quieter_via_method = sound1.apply_gain(-5.7)
+quieter_via_operator = sound1 - 5.7
+```
+
+### AudioSegment(…).fade()
+
+A more general (more flexible) fade method. You may specify `start` and `end`, or one of the two along with duration (e.g., `start` and `duration`).
+
+```python
+from pydub import AudioSegment
+sound1 = AudioSegment.from_file("sound1.wav")
+
+fade_louder_for_3_seconds_in_middle = sound1.fade(to_gain=+6.0, start=7500, duration=3000)
+
+fade_quieter_beteen_2_and_3_seconds = sound1.fade(to_gain=-3.5, start=2000, end=3000)
+
+# easy way is to use the .fade_in() convenience method. note: -120dB is basically silent.
+fade_in_the_hard_way = sound1.fade(from_gain=-120.0, start=0, duration=5000)
+fade_out_the_hard_way = sound1.fade(to_gain=-120.0, end=0, duration=5000)
+```
+
+**Supported keyword arguments**:
+
+- `to_gain` | example: `-3.0` | default: `0` (0dB, no change)
+  Resulting change at the end of the fade. `-6.0` means fade will be be from 0dB (no change) to -6dB, and everything after the fade will be -6dB.
+- `from_gain` | example: `-3.0` | default: `0` (0dB, no change)
+  Change at the beginning of the fade. `-6.0` means fade (and all audio before it) will be be at -6dB will fade up to 0dB – the rest of the audio after the fade will be at 0dB (i.e., unchanged).
+- `start` | example: `7500` | NO DEFAULT
+  Position to begin fading (in milliseconds). `5500` means fade will begin after 5.5 seconds.
+- `end` | example: `4` | NO DEFAULT
+  The overlaid `AudioSegment` will repeat X times (starting at `position`) but will still be truncated to the length of this `AudioSegment`
+- `duration` | example: `4` | NO DEFAULT
+  You can use `start` or `end` with duration, instead of specifying both - provided as a convenience.
+
+### AudioSegment(…).fade_out()
+
+Fade out (to silent) the end of this `AudioSegment`. Uses `.fade()` internally.
+
+**Supported keyword arguments**:
+
+- `duration` | example: `5000` | NO DEFAULT
+  How long (in milliseconds) the fade should last. Passed directly to `.fade()` internally
+
+### AudioSegment(…).fade_in()
+
+Fade in (from silent) the beginning of this `AudioSegment`. Uses `.fade()` internally.
+
+**Supported keyword arguments**:
+
+- `duration` | example: `5000` | NO DEFAULT
+  How long (in milliseconds) the fade should last. Passed directly to `.fade()` internally
+
+### AudioSegment(…).reverse()
+
+Make a copy of this `AudioSegment` that plays backwards. Useful for Pink Floyd, screwing around, and some audio processing algorithms.
+
+### AudioSegment(…).set_sample_width()
+
+Creates an equivalent version of this `AudioSegment` with the specified sample width (in bytes). Increasing this value does not generally cause a reduction in quality. Reducing it *definitely* does cause a loss in quality. Higher Sample width means more dynamic range.
+
+### AudioSegment(…).set_frame_rate()
+
+Creates an equivalent version of this `AudioSegment` with the specified frame rate (in Hz). Increasing this value does not generally cause a reduction in quality. Reducing it *definitely does* cause a loss in quality. Higher frame rate means larger frequency response (higher frequencies can be represented).
+
+### AudioSegment(…).set_channels()
+
+Creates an equivalent version of this `AudioSegment` with the specified number of channels (1 is Mono, 2 is Stereo). Converting from mono to stereo does not cause any audible change. Converting from stereo to mono may result in loss of quality (but only if the left and right chanels differ).
+
+### AudioSegment(…).split_to_mono()
+
+Splits a stereo `AudioSegment` into two, one for each channel (Left/Right). Returns a list with the new `AudioSegment` objects with the left channel at index 0 and the right channel at index 1.
+
+### AudioSegment(…).apply_gain_stereo()
+
+```python
+from pydub import AudioSegment
+sound1 = AudioSegment.from_file("sound1.wav")
+
+# make left channel 6dB quieter and right channe 2dB louder
+stereo_balance_adjusted = sound1.apply_gain_stereo(-6, +2)
+```
+Apply gain to the left and right channel of a stereo `AudioSegment`. If the `AudioSegment` is mono, it will be converted to stereo before applying the gain.
+
+Both gain arguments are specified in dB.
+
+### AudioSegment(…).pan()
+
+```python
+from pydub import AudioSegment
+sound1 = AudioSegment.from_file("sound1.wav")
+
+# pan the sound 15% to the right
+panned_right = sound1.pan(+0.15)
+
+# pan the sound 50% to the left
+panned_left = sound1.pan(-0.50)
+```
+
+Takes one positional argument, *pan amount*, which should be between -1.0 (100% left) and +1.0 (100% right)
+
+When pan_amount == 0.0 the left/right balance is not changed.
+
+Panning does not alter the *perceived* loundness, but since loudness
+is decreasing on one side, the other side needs to get louder to
+compensate. When panned hard left, the left channel will be 3dB louder and
+the right channel will be silent (and vice versa).
+
+### AudioSegment(…).get_array_of_samples()
+
+Returns the raw audio data as an array of (numeric) samples. Note: if the audio has multiple channels, the samples for each channel will be serialized – for example, stereo audio would look like `[sample_1_L, sample_1_R, sample_2_L, sample_2_R, …]`.
+
+This method is mainly for use in implementing effects, and other processing.
+
+```python
+from pydub import AudioSegment
+sound = AudioSegment.from_file(“sound1.wav”)
+
+samples = sound.get_array_of_samples()
+
+# then modify samples...
+
+new_sound = sound._spawn(samples)
+```
+
+note that when using numpy or scipy you will need to convert back to an array before you spawn:
+
+```python
+import array
+import numpy as np
+from pydub import AudioSegment
+
+sound = AudioSegment.from_file(“sound1.wav”)
+samples = sound.get_array_of_samples()
+
+# Example operation on audio data
+shifted_samples = np.right_shift(samples, 1)
+
+# now you have to convert back to an array.array
+shifted_samples_array = array.array(sound.array_type, shifted_samples)
+
+new_sound = sound._spawn(shifted_samples_array)
+```
+
+Here's how to convert to a numpy float32 array:
+
+```python
+import numpy as np
+from pydub import AudioSegment
+
+sound = AudioSegment.from_file("sound1.wav")
+sound = sound.set_frame_rate(16000)
+channel_sounds = sound.split_to_mono()
+samples = [s.get_array_of_samples() for s in channel_sounds]
+
+fp_arr = np.array(samples).T.astype(np.float32)
+fp_arr /= np.iinfo(samples[0].typecode).max
+```
+
+And how to convert it back to an AudioSegment:
+
+```python
+import io
+import scipy.io.wavfile
+
+wav_io = io.BytesIO()
+scipy.io.wavfile.write(wav_io, 16000, fp_arr)
+wav_io.seek(0)
+sound = pydub.AudioSegment.from_wav(wav_io)
+```
+
+### AudioSegment(…).get_dc_offset()
+
+Returns a value between -1.0 and 1.0 representing the DC offset of a channel. This is calculated using `audioop.avg()` and normalizing the result by samples max value.
+
+**Supported keyword arguments**:
+
+- `channel` | example: `2` | default: `1`
+  Selects left (1) or right (2) channel to calculate DC offset. If segment is mono, this value is ignored.
+
+### AudioSegment(…).remove_dc_offset()
+
+Removes DC offset from channel(s). This is done by using `audioop.bias()`, so watch out for overflows.
+
+**Supported keyword arguments**:
+
+- `channel` | example: `2` | default: None
+  Selects left (1) or right (2) channel remove DC offset. If value if None, removes from all available channels. If segment is mono, this value is ignored.
+
+- `offset` | example: `-0.1` | default: None
+  Offset to be removed from channel(s). Calculates offset if it's None. Offset values must be between -1.0 and 1.0.
+
+## Effects
+
+Collection of DSP effects that are implemented by `AudioSegment` objects.
+
+### AudioSegment(…).invert_phase()
+
+Make a copy of this `AudioSegment` and inverts the phase of the signal. Can generate anti-phase waves for noise suppression or cancellation.
+
+## Silence
+
+Various functions for finding/manipulating silence in AudioSegments. For creating silent AudioSegments, see AudioSegment.silent().
+
+### silence.detect_silence()
+
+Returns a list of all silent sections [start, end] in milliseconds of audio_segment. Inverse of detect_nonsilent(). Can be very slow since it has to iterate over the whole segment.
+
+```python
+from pydub import AudioSegment, silence
+
+print(silence.detect_silence(AudioSegment.silent(2000)))
+# [[0, 2000]]
+```
+
+**Supported keyword arguments**:
+
+- `min_silence_len` | example: `500` | default: 1000
+  The minimum length for silent sections in milliseconds. If it is greater than the length of the audio segment an empty list will be returned.
+
+- `silence_thresh` | example: `-20` | default: -16
+  The upper bound for how quiet is silent in dBFS.
+
+- `seek_step` | example: `5` | default: 1
+  Size of the step for checking for silence in milliseconds. Smaller is more precise. Must be a positive whole number.
+
+### silence.detect_nonsilent()
+
+Returns a list of all silent sections [start, end] in milliseconds of audio_segment. Inverse of detect_silence() and has all the same arguments. Can be very slow since it has to iterate over the whole segment.
+
+**Supported keyword arguments**:
+
+- `min_silence_len` | example: `500` | default: 1000
+  The minimum length for silent sections in milliseconds. If it is greater than the length of the audio segment an empty list will be returned.
+
+- `silence_thresh` | example: `-20` | default: -16
+  The upper bound for how quiet is silent in dBFS.
+
+- `seek_step` | example: `5` | default: 1
+  Size of the step for checking for silence in milliseconds. Smaller is more precise. Must be a positive whole number.
+
+### silence.split_on_silence()
+
+Returns list of audio segments from splitting audio_segment on silent sections.
+
+**Supported keyword arguments**:
+
+- `min_silence_len` | example: `500` | default: 1000
+  The minimum length for silent sections in milliseconds. If it is greater than the length of the audio segment an empty list will be returned.
+
+- `silence_thresh` | example: `-20` | default: -16
+  The upper bound for how quiet is silent in dBFS.
+
+- `seek_step` | example: `5` | default: 1
+  Size of the step for checking for silence in milliseconds. Smaller is more precise. Must be a positive whole number.
+
+- `keep_silence` ~ example: True | default: 100
+  How much silence to keep in ms or a bool. leave some silence at the beginning and end of the chunks. Keeps the sound from sounding like it is abruptly cut off.
+  When the length of the silence is less than the keep_silence duration it is split evenly between the preceding and following non-silent segments.
+  If True is specified, all the silence is kept, if False none is kept.
+
+### silence.detect_leading_silence()
+
+Returns the millisecond/index that the leading silence ends. If there is no end it will return the length of the audio_segment.
+
+```python
+from pydub import AudioSegment, silence
+
+print(silence.detect_silence(AudioSegment.silent(2000)))
+# 2000
+```
+
+**Supported keyword arguments**:
+
+- `silence_thresh` | example: `-20` | default: -50
+  The upper bound for how quiet is silent in dBFS.
+
+- `chunk_size` | example: `5` | default: 10
+  Size of the step for checking for silence in milliseconds. Smaller is more precise. Must be a positive whole number.
@@ -0,0 +1,101 @@
+James Robert
+    github: jiaaro
+    twitter: @jiaaro
+    web: jiaaro.com
+    email: pydub@jiaaro.com
+
+Marc Webbie
+    github: marcwebbie
+
+Jean-philippe Serafin
+    github: jeanphix
+
+Anurag Ramdasan
+    github: AnuragRamdasan
+
+Choongmin Lee
+    github: clee704
+
+Patrick Pittman
+    github: ptpittman
+
+Hunter Lang
+    github: hunterlang
+
+Alexey
+    github: nihisil
+
+Jaymz Campbell
+    github: jaymzcd
+
+Ross McFarland
+    github: ross
+
+John McMellen
+    github: jmcmellen
+
+Johan Lövgren
+    github: dashj
+
+Joachim Krüger
+    github: jkrgr
+
+Shichao An
+    github: shichao-an
+
+Michael Bortnyck
+    github: mbortnyck
+
+André Cloete
+    github: aj-cloete
+
+David Acacio
+    github: dacacioa
+
+Thiago Abdnur
+    github: bolaum
+
+Aurélien Ooms
+    github: aureooms
+
+Mike Mattozzi
+    github: mmattozzi
+
+Marcio Mazza
+    github: marciomazza
+
+Sungsu Lim
+    github: proflim
+
+Evandro Myller
+    github: emyller
+
+Sérgio Agostinho
+    github: SergioRAgostinho
+
+Antonio Larrosa
+    github: antlarr
+
+Aaron Craig
+    github: craigthelinguist
+
+Carlos del Castillo
+    github: greyalien502
+
+Yudong Sun
+    github: sunjerry019
+
+Jorge Perianez
+    github: JPery
+
+Chendi Luo
+    github: Creonalia
+
+Daniel Lefevre
+    gitHub: dplefevre
+
+Grzegorz Kotfis
+    github: gkotfis
+
+Pål Orby
+    github: orby
@@ -0,0 +1,168 @@
+# v0.25.1
+- Fix crashing bug in new scipy-powered EQ effects
+
+# v0.25.0
+- Don't show a runtime warning about the optional ffplay dependency being missing until someone trys to use it
+- Documentation improvements
+- Python 3.9 support
+- Improved efficiency of loading wave files with `pydub.AudioSegment.from_file()`
+- Ensure `pydub.AudioSegment().export()` always retuns files with a seek position at the beginning of the file   
+- Added more EQ effects to `pydub.scipy_effects` (requires scipy to be installed)
+- Fix a packaging bug where the LICENSE file was not included in the source distribution
+- Add a way to instantiate a `pydub.AudioSegment()` with a portion of an audio file via `pydub.AudioSegment().from_file()`
+
+# v0.24.1
+- Fix bug where ffmpeg errors in Python 3 are illegible
+- Fix bug where `split_on_silence` fails when there are one or fewer nonsilent segments
+- Fix bug in fallback audioop implementation
+
+# v0.24.0
+- Fix inconsistent handling of 8-bit audio
+- Fix bug where certain files will fail to parse
+- Fix bug where pyaudio stream is not closed on error
+- Allow codecs and parameters in wav and raw export
+- Fix bug in `pydub.AudioSegment.from_file` where supplied codec is ignored
+- Allow `pydub.silence.split_on_silence` to take a boolean for `keep_silence`
+- Fix bug where `pydub.silence.split_on_silence` sometimes adds non-silence from adjacent segments
+- Fix bug where `pydub.AudioSegment.extract_wav_headers` fails on empty wav files
+- Add new function `pydub.silence.detect_leading_silence`
+- Support conversion between an arbitrary number of channels and mono in `pydub.AudioSegment.set_channels`
+- Fix several issues related to reading from filelike objects
+
+# v0.23.1
+- Fix bug in passing ffmpeg/avconv parameters for `pydub.AudioSegment.from_mp3()`, `pydub.AudioSegment.from_flv()`, `pydub.AudioSegment.from_ogg()`, and `pydub.AudioSegment.from_wav()`
+- Fix logic bug in `pydub.effects.strip_silence()`
+
+# v0.23.0
+- Add support for playback via simpleaudio
+- Allow users to override the type in `pydub.AudioSegment().get_array_of_samples()` (PR #313)
+- Fix a bug where the wrong codec was used for 8-bit audio (PR #309 - issue #308)
+
+# v0.22.1
+- Fix `pydub.utils.mediainfo_json()` to work with newer, backwards-incompatible versions of ffprobe/avprobe
+
+# v0.22.0
+- Adds support for audio with frame rates (sample rates) of 48k and higher (requires scipy) (PR #262, fixes #134, #237, #209)
+- Adds support for PEP 519 File Path protocol (PR #252)
+- Fixes a few places where handles to temporary files are kept open (PR #280)
+- Add the license file to the python package to aid other packaging projects (PR #279, fixes #274)
+- Big fix for `pydub.silence.detect_silence()` (PR #263)
+
+# v0.21.0
+- NOTE: Semi-counterintuitive change: using the a stride when slicing AudioSegment instances (for example, `sound[::5000]`) will return chunks of 5000ms (not 1ms chunks every 5000ms) (#222)
+- Debug output from ffmpeg/avlib is no longer printed to the console unless you set up logging (see README for how to set up logging for your converter) (#223)
+- All pydub exceptions are now subclasses of `pydub.exceptions.PydubException` (PR #244)
+- The utilities in `pydub.silence` now accept a `seek_step`argument which can optionally be passed to improve the performance of silence detection (#211)
+- Fix to `pydub.silence` utilities which allow you to detect perfect silence (#233)
+- Fix a bug where threaded code screws up your terminal session due to ffmpeg inheriting the stdin from the parent process. (#231)
+- Fix a bug where a crashing programs using pydub would leave behind their temporary files (#206)
+
+# v0.20.0
+- Add new parameter `gain_during_overlay` to `pydub.AudioSegment.overlay` which allows users to adjust the volume of the target AudioSegment during the portion of the segment which is overlaid with the additional AudioSegment.
+- `pydub.playback.play()` No longer displays the (very verbose) playback "banner" when using ffplay
+- Fix a confusing error message when using invalid crossfade durations (issue #193)
+
+# v0.19.0
+- Allow codec and ffmpeg/avconv parameters to be set in the `pydub.AudioSegment.from_file()` for more control while decoding audio files
+- Allow `AudioSegment` objects with more than two channels to be split using `pydub.AudioSegment().split_to_mono()`
+- Add support for inverting the phase of only one channel in a multi-channel `pydub.AudioSegment` object
+- Fix a bug with the latest avprobe that broke `pydub.utils.mediainfo()`
+- Add tests for webm encoding/decoding
+
+# v0.18.0
+- Add a new constructor: `pydub.AudioSegment.from_mono_audiosegments()` which allows users to create a multi-channel audiosegment out of multiple mono ones.
+- Refactor `pydub.AudioSegment._sync()` to support an arbitrary number of audiosegment arguments.
+
+# v0.17.0
+- Add the ability to add a cover image to MP3 exports via the `cover` keyword argument to `pydub.AudioSegment().export()`
+- Add `pydub.AudioSegment().get_dc_offset()` and `pydub.AudioSegment().remove_dc_offset()` which allow detection and removal of DC offset in audio files.
+- Minor fixes for windows users
+
+# v0.16.7
+- Make `pydub.AudioSegment()._spawn()` accept array.array instances containing audio samples
+
+# v0.16.6
+- Make `pydub.AudioSegment()` objects playable inline in ipython notebooks.
+- Add scipy powered high pass, low pass, and band pass filters, which can be high order filters (they take `order` as a keyword argument). They are used for `pydub.AudioSegment().high_pass_filter()`, `pydub.AudioSegment().low_pass_filter()`, `pydub.AudioSegment().band_pass_filter()` when the `pydub.scipy_effects` module is imported.
+- Fix minor bug in `pydub.silence.detect_silence()`
+
+# v0.16.5
+- Update `pydub.AudioSegment()._spawn()` method to allow user subclassing of `pydub.AudioSegment`
+- Add a workaround for incorrect duration reporting of some mp3 files on macOS
+
+# v0.16.4
+- Add support for radd (basically, allow `sum()` to operate on an iterable of `pydub.AudioSegment()` objects)
+- Fix bug in 24-bit wav support (understatement. It didn't work right at all the first time)
+
+# v0.16.3
+- Add support for python 3.5 (overstatement. We just added python 3.5 to CI and it worked 😄)
+- Add native support for 24-bit wav files (ffmpeg/avconv not required)
+
+# v0.16.2
+- Fix bug where you couldn't directly instantiate `pydub.AudioSegment` with `bytes` data in python 3
+
+# v0.16.1
+- pydub will use any ffmpeg/avconv binary that's in the current directory (as reported by `os.getcwd()`) before searching for a system install
+
+# v0.16.0
+- Make it easier to instantiate `pydub.AudioSegment()` directly when creating audio segments from raw audio data (without having to write it to a file first)
+- Add `pydub.AudioSegment().get_array_of_samples()` method which returns the samples which make up an audio segment (you should usually prefer this over `pydub.AudioSegment().raw_data`)
+- Add `pydub.AudioSegment().raw_data` property which returns the raw audio data for an audio segment as a bytes (python 3) or a bytestring (python 3)
+- Allow users to specify frame rate in `pydub.AudioSegment.silent()` constructor
+
+# v0.15.0
+- Add support for RAW audio (basically WAV format, but without wave headers)
+- Add a new exception `pydub.exceptions.CouldntDecodeError` to indicate a failure of ffmpeg/avconv to decode a file (as indicated by ffmpeg/avconv exit code)
+
+# v0.14.2
+- Fix a bug in python 3.4 which failed to read wave files with no audio data (should have been audio segments with a duration of 0 ms)
+
+# v0.14.1
+- Fix a bug in `pydub.utils.mediainfo()` that caused inputs containing unescaped characters to raise a runtime error (inputs are not supposed to require escaping)
+
+# v0.14.0
+- Rename `pydub.AudioSegment().set_gain()` to `pydub.AudioSegment().apply_gain_stereo()` to better reflect it's place in the world (as a counterpart to `pydub.AudioSegment().apply_gain()`)
+
+# v0.13.0
+- Add `pydub.AudioSegment().pan()` which returns a new stereo audio segment panned left/right as specified.
+
+# v0.12.0
+- Add a logger, `"pydub.converter"` which logs the ffmpeg commands being run by pydub.
+- Add `pydub.AudioSegment().split_to_mono()` method which returns a list of mono audio segments. One for each channel in the original audio segment.
+- Fix a bug in `pydub.silence.detect_silence()` which caused the function to break when a silent audio segment was equal in length to the minimum silence length. It should report a single span of silence covering the whole silent audio segment. Now it does.
+- Fix a bug where uncommon wav formats (those not supported by the stdlib wave module) would throw an exception rather than converting to a more common format via ffmpeg/avconv
+
+# v0.11.0
+- Add `pydub.AudioSegment().max_dBFS` which reports the loudness (in dBFS) of the loudest point (i.e., highest amplitude sample) of an audio segment
+
+# v0.10.0
+- Overhaul Documentation
+- Improve performance of `pydub.AudioSegment().overlay()`
+- Add `pydub.AudioSegment().invert_phase()` which (shocker) inverts the phase of an audio segment
+- Fix a type error in `pydub.AudioSegment.get_sample_slice()`
+
+# v0.9.5
+- Add `pydub.generators` module containing simple signal generation functions (white noise, sine, square wave, etc)
+- Add a `loops` keyword argument to `pydub.AudioSegment().overlay()` which allows users to specify that the overlaid audio should be repeated (i.e., looped) a certain number of times, or indefinitely
+
+# 0.9.4
+- Fix a bug in db_to_float() where all values were off by a factor of 2
+
+# 0.9.3
+- Allow users to set the location of their converter by setting `pydub.AudioSegment.converter = "/path/to/ffmpeg"` and added a shim to support the old method of assigning to `pydub.AudioSegment.ffmpeg` (which is deprecated now that we support avconv)
+
+# v0.9.2
+- Add support for Python 3.4
+- Audio files opened with format "wave" are treated as "wav" and "m4a" are treated as "mp4"
+- Add `pydub.silence` module with simple utilities for detecting and removing silence.
+- Fix a bug affecting auto-detection of ffmpeg/avconv on windows.
+- Fix a bug that caused pydub to only work when ffmpeg/avconv is present (it should be able to work with WAV data without any dependencies)
+
+# v0.9.1
+- Add a runtime warning when ffmpeg/avconv cannot be found to aid debugging
+
+# v0.9.0
+- Added support for pypy (by reimplementing audioop in python). Also, we contributed our implementation to the pypy project, so that's 💯
+- Add support for avconv as an alternative to ffmpeg
+- Add a new helper module `pydub.playback` which allows you to quickly listen to an audio segment using ffplay (or avplay)
+- Add new function `pydub.utils.mediainfo('/path/to/audio/file.ext')` which reports back the results of ffprobe (or avprobe) including codec, bitrate, channels, etc
@@ -0,0 +1,45 @@
+Pydub loves user contributions. 
+
+We are happy to merge Pull Requests for features and bug fixes, of course. But, also spelling corrections, PEP 8 conformity, and platform-specific fixes.
+
+Don't be shy!
+
+### How to contribute:
+
+1. Fork [pydub on github](https://github.com/jiaaro/pydub)
+2. Commit changes
+3. Send a Pull Request
+
+you did it! 
+
+don't forget to append your name to the AUTHORS file ;)
+
+There _are_ a few things that will make your Pull Request more likely to be merged:
+
+1. Maintain backward compatibility
+2. Avoid new dependencies
+3. Include tests (and make sure they pass)
+4. Write a short description of **what** is changed and **why**
+5. Keep your Pull Request small, and focused on fixing one thing. 
+
+  Smaller is easier to review, and easier to understand. 
+  
+  If you want to fix spelling and PEP 8 violations, send two pull requests :)
+  
+  
+### Want to pitch in?
+  
+Take a look at our issue tracker for anything tagged [`bug`][bugs] or [`todo`][todos] - these are goals of the project and your improvements are _very_ likely to be merged!
+
+That being said, there are many possible contributions we haven't thought of already. Those are welcome too!
+
+Here are some general topics of interest for future development:
+
+- Make it easier to get started with pydub
+- More/better audio effects
+- Support more audio formats
+- Improve handling of large audio files
+- Make things faster and use less memory.
+
+[bugs]: https://github.com/jiaaro/pydub/issues?q=is%3Aissue+is%3Aopen+label%3Abug
+[todos]: https://github.com/jiaaro/pydub/issues?q=is%3Aissue+is%3Aopen+label%3Atodo
@@ -0,0 +1,20 @@
+Copyright (c) 2011 James Robert, http://jiaaro.com
+
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice shall be
+included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
+LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
@@ -0,0 +1 @@
+include LICENSE
@@ -0,0 +1,333 @@
+# Pydub [![Build Status](https://travis-ci.org/jiaaro/pydub.svg?branch=master)](https://travis-ci.org/jiaaro/pydub) [![Build status](https://ci.appveyor.com/api/projects/status/gy1ucp9o5khq7fqi/branch/master?svg=true)](https://ci.appveyor.com/project/jiaaro/pydub/branch/master)
+
+Pydub lets you do stuff to audio in a way that isn't stupid.
+
+**Stuff you might be looking for**:
+ - [Installing Pydub](https://github.com/jiaaro/pydub#installation)
+ - [API Documentation](https://github.com/jiaaro/pydub/blob/master/API.markdown)
+ - [Dependencies](https://github.com/jiaaro/pydub#dependencies)
+ - [Playback](https://github.com/jiaaro/pydub#playback)
+ - [Setting up ffmpeg](https://github.com/jiaaro/pydub#getting-ffmpeg-set-up)
+ - [Questions/Bugs](https://github.com/jiaaro/pydub#bugs--questions)
+ 
+
+##  Quickstart
+
+Open a WAV file
+
+```python
+from pydub import AudioSegment
+
+song = AudioSegment.from_wav("never_gonna_give_you_up.wav")
+```
+
+...or a mp3
+
+```python
+song = AudioSegment.from_mp3("never_gonna_give_you_up.mp3")
+```
+
+... or an ogg, or flv, or [anything else ffmpeg supports](http://www.ffmpeg.org/general.html#File-Formats)
+
+```python
+ogg_version = AudioSegment.from_ogg("never_gonna_give_you_up.ogg")
+flv_version = AudioSegment.from_flv("never_gonna_give_you_up.flv")
+
+mp4_version = AudioSegment.from_file("never_gonna_give_you_up.mp4", "mp4")
+wma_version = AudioSegment.from_file("never_gonna_give_you_up.wma", "wma")
+aac_version = AudioSegment.from_file("never_gonna_give_you_up.aiff", "aac")
+```
+
+Slice audio:
+
+```python
+# pydub does things in milliseconds
+ten_seconds = 10 * 1000
+
+first_10_seconds = song[:ten_seconds]
+
+last_5_seconds = song[-5000:]
+```
+
+Make the beginning louder and the end quieter
+
+```python
+# boost volume by 6dB
+beginning = first_10_seconds + 6
+
+# reduce volume by 3dB
+end = last_5_seconds - 3
+```
+
+Concatenate audio (add one file to the end of another)
+
+```python
+without_the_middle = beginning + end
+```
+
+How long is it?
+
+```python
+without_the_middle.duration_seconds == 15.0
+```
+
+AudioSegments are immutable
+
+```python
+# song is not modified
+backwards = song.reverse()
+```
+
+Crossfade (again, beginning and end are not modified)
+
+```python
+# 1.5 second crossfade
+with_style = beginning.append(end, crossfade=1500)
+```
+
+Repeat
+
+```python
+# repeat the clip twice
+do_it_over = with_style * 2
+```
+
+Fade (note that you can chain operations because everything returns
+an AudioSegment)
+
+```python
+# 2 sec fade in, 3 sec fade out
+awesome = do_it_over.fade_in(2000).fade_out(3000)
+```
+
+Save the results (again whatever ffmpeg supports)
+
+```python
+awesome.export("mashup.mp3", format="mp3")
+```
+
+Save the results with tags (metadata)
+
+```python
+awesome.export("mashup.mp3", format="mp3", tags={'artist': 'Various artists', 'album': 'Best of 2011', 'comments': 'This album is awesome!'})
+```
+
+You can pass an optional bitrate argument to export using any syntax ffmpeg 
+supports.
+
+```python
+awesome.export("mashup.mp3", format="mp3", bitrate="192k")
+```
+
+Any further arguments supported by ffmpeg can be passed as a list in a 
+'parameters' argument, with switch first, argument second. Note that no 
+validation takes place on these parameters, and you may be limited by what 
+your particular build of ffmpeg/avlib supports.
+
+```python
+# Use preset mp3 quality 0 (equivalent to lame V0)
+awesome.export("mashup.mp3", format="mp3", parameters=["-q:a", "0"])
+
+# Mix down to two channels and set hard output volume
+awesome.export("mashup.mp3", format="mp3", parameters=["-ac", "2", "-vol", "150"])
+```
+
+## Debugging
+
+Most issues people run into are related to converting between formats using
+ffmpeg/avlib. Pydub provides a logger that outputs the subprocess calls to 
+help you track down issues:
+
+```python
+>>> import logging
+
+>>> l = logging.getLogger("pydub.converter")
+>>> l.setLevel(logging.DEBUG)
+>>> l.addHandler(logging.StreamHandler())
+
+>>> AudioSegment.from_file("./test/data/test1.mp3")
+subprocess.call(['ffmpeg', '-y', '-i', '/var/folders/71/42k8g72x4pq09tfp920d033r0000gn/T/tmpeZTgMy', '-vn', '-f', 'wav', '/var/folders/71/42k8g72x4pq09tfp920d033r0000gn/T/tmpK5aLcZ'])
+<pydub.audio_segment.AudioSegment object at 0x101b43e10>
+```
+
+Don't worry about the temporary files used in the conversion. They're cleaned up 
+automatically.
+
+## Bugs & Questions
+
+You can file bugs in our [github issues tracker](https://github.com/jiaaro/pydub/issues), 
+and ask any technical questions on 
+[Stack Overflow using the pydub tag](http://stackoverflow.com/questions/ask?tags=pydub). 
+We keep an eye on both.
+
+## Installation
+
+Installing pydub is easy, but don't forget to install ffmpeg/avlib (the next section in this doc)
+
+    pip install pydub
+
+Or install the latest dev version from github (or replace `@master` with a [release version like `@v0.12.0`](https://github.com/jiaaro/pydub/releases))…
+
+    pip install git+https://github.com/jiaaro/pydub.git@master
+
+-OR-
+
+    git clone https://github.com/jiaaro/pydub.git
+
+-OR-
+
+Copy the pydub directory into your python path. Zip 
+[here](https://github.com/jiaaro/pydub/zipball/master)
+
+## Dependencies
+
+You can open and save WAV files with pure python. For opening and saving non-wav 
+files – like mp3 – you'll need [ffmpeg](http://www.ffmpeg.org/) or 
+[libav](http://libav.org/).
+
+### Playback
+
+You can play audio if you have one of these installed (simpleaudio _strongly_ recommended, even if you are installing ffmpeg/libav):
+
+ - [simpleaudio](https://simpleaudio.readthedocs.io/en/latest/)
+ - [pyaudio](https://people.csail.mit.edu/hubert/pyaudio/docs/#)
+ - ffplay (usually bundled with ffmpeg, see the next section)
+ - avplay (usually bundled with libav, see the next section)
+ 
+```python
+from pydub import AudioSegment
+from pydub.playback import play
+
+sound = AudioSegment.from_file("mysound.wav", format="wav")
+play(sound)
+```
+
+## Getting ffmpeg set up
+
+You may use **libav or ffmpeg**.
+
+Mac (using [homebrew](http://brew.sh)):
+
+```bash
+# libav
+brew install libav
+
+####    OR    #####
+
+# ffmpeg
+brew install ffmpeg
+```
+
+Linux (using aptitude):
+
+```bash
+# libav
+apt-get install libav-tools libavcodec-extra
+
+####    OR    #####
+
+# ffmpeg
+apt-get install ffmpeg libavcodec-extra
+```
+
+Windows:
+
+1. Download and extract libav from [Windows binaries provided here](http://builds.libav.org/windows/).
+2. Add the libav `/bin` folder to your PATH envvar
+3. `pip install pydub`
+
+## Important Notes
+
+`AudioSegment` objects are [immutable](http://www.devshed.com/c/a/Python/String-and-List-Python-Object-Types/1/)
+
+
+### Ogg exporting and default codecs
+
+The Ogg specification ([http://tools.ietf.org/html/rfc5334](rfc5334)) does not specify
+the codec to use, this choice is left up to the user. Vorbis and Theora are just
+some of a number of potential codecs (see page 3 of the rfc) that can be used for the
+encapsulated data.
+
+When no codec is specified exporting to `ogg` will _default_ to using `vorbis`
+as a convenience. That is:
+
+```python
+from pydub import AudioSegment
+song = AudioSegment.from_mp3("test/data/test1.mp3")
+song.export("out.ogg", format="ogg")  # Is the same as:
+song.export("out.ogg", format="ogg", codec="libvorbis")
+```
+
+## Example Use
+
+Suppose you have a directory filled with *mp4* and *flv* videos and you want to convert all of them to *mp3* so you can listen to  them on your mp3 player.
+
+```python
+import os
+import glob
+from pydub import AudioSegment
+
+video_dir = '/home/johndoe/downloaded_videos/'  # Path where the videos are located
+extension_list = ('*.mp4', '*.flv')
+
+os.chdir(video_dir)
+for extension in extension_list:
+    for video in glob.glob(extension):
+        mp3_filename = os.path.splitext(os.path.basename(video))[0] + '.mp3'
+        AudioSegment.from_file(video).export(mp3_filename, format='mp3')
+```
+
+### How about another example?
+
+```python
+from glob import glob
+from pydub import AudioSegment
+
+playlist_songs = [AudioSegment.from_mp3(mp3_file) for mp3_file in glob("*.mp3")]
+
+first_song = playlist_songs.pop(0)
+
+# let's just include the first 30 seconds of the first song (slicing
+# is done by milliseconds)
+beginning_of_song = first_song[:30*1000]
+
+playlist = beginning_of_song
+for song in playlist_songs:
+
+    # We don't want an abrupt stop at the end, so let's do a 10 second crossfades
+    playlist = playlist.append(song, crossfade=(10 * 1000))
+
+# let's fade out the end of the last song
+playlist = playlist.fade_out(30)
+
+# hmm I wonder how long it is... ( len(audio_segment) returns milliseconds )
+playlist_length = len(playlist) / (1000*60)
+
+# lets save it!
+with open("%s_minute_playlist.mp3" % playlist_length, 'wb') as out_f:
+    playlist.export(out_f, format='mp3')
+```
+
+## License ([MIT License](http://opensource.org/licenses/mit-license.php))
+
+Copyright © 2011 James Robert, http://jiaaro.com
+
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice shall be
+included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
+LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
@@ -0,0 +1,30 @@
+build: false
+environment:
+  matrix:
+    - PYTHON: "C:/Python27"
+      FFMPEG: "4.2.3"
+    - PYTHON: "C:/Python34"
+      FFMPEG: "4.2.3"
+    - PYTHON: "C:/Python35"
+      FFMPEG: "4.2.3"
+    - PYTHON: "C:/Python36"
+      FFMPEG: "4.2.3"
+    - PYTHON: "C:/Python36"
+      FFMPEG: "latest"
+matrix:
+  allow_failures:
+    - FFMPEG: "latest"
+init:
+  - "ECHO %PYTHON%"
+  - ps: "ls C:/Python*"
+install:
+  - "%PYTHON%/python.exe -m pip install wheel"
+  - "%PYTHON%/python.exe -m pip install -e ."
+  # Install ffmpeg
+  - ps: Start-FileDownload ('https://github.com/advancedfx/ffmpeg.zeranoe.com-builds-mirror/releases/download/20200915/ffmpeg-' + $env:FFMPEG + '-win64-shared.zip') ffmpeg-shared.zip
+  - 7z x ffmpeg-shared.zip > NULL
+  - "SET PATH=%cd%\\ffmpeg-%FFMPEG%-win64-shared\\bin;%PATH%"
+  # check ffmpeg installation (also shows version)
+  - "ffmpeg.exe -version"
+test_script:
+  - "%PYTHON%/python.exe test/test.py"
@@ -0,0 +1 @@
+from .audio_segment import AudioSegment
@@ -0,0 +1,341 @@
+import sys
+import math
+import array
+from .utils import (
+    db_to_float,
+    ratio_to_db,
+    register_pydub_effect,
+    make_chunks,
+    audioop,
+    get_min_max_value
+)
+from .silence import split_on_silence
+from .exceptions import TooManyMissingFrames, InvalidDuration
+
+if sys.version_info >= (3, 0):
+    xrange = range
+
+
+@register_pydub_effect
+def apply_mono_filter_to_each_channel(seg, filter_fn):
+    n_channels = seg.channels
+
+    channel_segs = seg.split_to_mono()
+    channel_segs = [filter_fn(channel_seg) for channel_seg in channel_segs]
+
+    out_data = seg.get_array_of_samples()
+    for channel_i, channel_seg in enumerate(channel_segs):
+        for sample_i, sample in enumerate(channel_seg.get_array_of_samples()):
+            index = (sample_i * n_channels) + channel_i
+            out_data[index] = sample
+
+    return seg._spawn(out_data)
+
+
+@register_pydub_effect
+def normalize(seg, headroom=0.1):
+    """
+    headroom is how close to the maximum volume to boost the signal up to (specified in dB)
+    """
+    peak_sample_val = seg.max
+    
+    # if the max is 0, this audio segment is silent, and can't be normalized
+    if peak_sample_val == 0:
+        return seg
+    
+    target_peak = seg.max_possible_amplitude * db_to_float(-headroom)
+
+    needed_boost = ratio_to_db(target_peak / peak_sample_val)
+    return seg.apply_gain(needed_boost)
+
+
+@register_pydub_effect
+def speedup(seg, playback_speed=1.5, chunk_size=150, crossfade=25):
+    # we will keep audio in 150ms chunks since one waveform at 20Hz is 50ms long
+    # (20 Hz is the lowest frequency audible to humans)
+
+    # portion of AUDIO TO KEEP. if playback speed is 1.25 we keep 80% (0.8) and
+    # discard 20% (0.2)
+    atk = 1.0 / playback_speed
+
+    if playback_speed < 2.0:
+        # throwing out more than half the audio - keep 50ms chunks
+        ms_to_remove_per_chunk = int(chunk_size * (1 - atk) / atk)
+    else:
+        # throwing out less than half the audio - throw out 50ms chunks
+        ms_to_remove_per_chunk = int(chunk_size)
+        chunk_size = int(atk * chunk_size / (1 - atk))
+
+    # the crossfade cannot be longer than the amount of audio we're removing
+    crossfade = min(crossfade, ms_to_remove_per_chunk - 1)
+
+    # DEBUG
+    #print("chunk: {0}, rm: {1}".format(chunk_size, ms_to_remove_per_chunk))
+
+    chunks = make_chunks(seg, chunk_size + ms_to_remove_per_chunk)
+    if len(chunks) < 2:
+        raise Exception("Could not speed up AudioSegment, it was too short {2:0.2f}s for the current settings:\n{0}ms chunks at {1:0.1f}x speedup".format(
+            chunk_size, playback_speed, seg.duration_seconds))
+
+    # we'll actually truncate a bit less than we calculated to make up for the
+    # crossfade between chunks
+    ms_to_remove_per_chunk -= crossfade
+
+    # we don't want to truncate the last chunk since it is not guaranteed to be
+    # the full chunk length
+    last_chunk = chunks[-1]
+    chunks = [chunk[:-ms_to_remove_per_chunk] for chunk in chunks[:-1]]
+
+    out = chunks[0]
+    for chunk in chunks[1:]:
+        out = out.append(chunk, crossfade=crossfade)
+
+    out += last_chunk
+    return out
+    
+
+@register_pydub_effect
+def strip_silence(seg, silence_len=1000, silence_thresh=-16, padding=100):
+    if padding > silence_len:
+        raise InvalidDuration("padding cannot be longer than silence_len")
+
+    chunks = split_on_silence(seg, silence_len, silence_thresh, padding)
+    crossfade = padding / 2
+
+    if not len(chunks):
+        return seg[0:0]
+
+    seg = chunks[0]
+    for chunk in chunks[1:]:
+        seg = seg.append(chunk, crossfade=crossfade)
+
+    return seg
+
+
+@register_pydub_effect
+def compress_dynamic_range(seg, threshold=-20.0, ratio=4.0, attack=5.0, release=50.0):
+    """
+    Keyword Arguments:
+        
+        threshold - default: -20.0
+            Threshold in dBFS. default of -20.0 means -20dB relative to the
+            maximum possible volume. 0dBFS is the maximum possible value so
+            all values for this argument sould be negative.
+
+        ratio - default: 4.0
+            Compression ratio. Audio louder than the threshold will be 
+            reduced to 1/ratio the volume. A ratio of 4.0 is equivalent to
+            a setting of 4:1 in a pro-audio compressor like the Waves C1.
+        
+        attack - default: 5.0
+            Attack in milliseconds. How long it should take for the compressor
+            to kick in once the audio has exceeded the threshold.
+
+        release - default: 50.0
+            Release in milliseconds. How long it should take for the compressor
+            to stop compressing after the audio has falled below the threshold.
+
+    
+    For an overview of Dynamic Range Compression, and more detailed explanation
+    of the related terminology, see: 
+
+        http://en.wikipedia.org/wiki/Dynamic_range_compression
+    """
+
+    thresh_rms = seg.max_possible_amplitude * db_to_float(threshold)
+    
+    look_frames = int(seg.frame_count(ms=attack))
+    def rms_at(frame_i):
+        return seg.get_sample_slice(frame_i - look_frames, frame_i).rms
+    def db_over_threshold(rms):
+        if rms == 0: return 0.0
+        db = ratio_to_db(rms / thresh_rms)
+        return max(db, 0)
+
+    output = []
+
+    # amount to reduce the volume of the audio by (in dB)
+    attenuation = 0.0
+    
+    attack_frames = seg.frame_count(ms=attack)
+    release_frames = seg.frame_count(ms=release)
+    for i in xrange(int(seg.frame_count())):
+        rms_now = rms_at(i)
+        
+        # with a ratio of 4.0 this means the volume will exceed the threshold by
+        # 1/4 the amount (of dB) that it would otherwise
+        max_attenuation = (1 - (1.0 / ratio)) * db_over_threshold(rms_now)
+        
+        attenuation_inc = max_attenuation / attack_frames
+        attenuation_dec = max_attenuation / release_frames
+        
+        if rms_now > thresh_rms and attenuation <= max_attenuation:
+            attenuation += attenuation_inc
+            attenuation = min(attenuation, max_attenuation)
+        else:
+            attenuation -= attenuation_dec
+            attenuation = max(attenuation, 0)
+        
+        frame = seg.get_frame(i)
+        if attenuation != 0.0:
+            frame = audioop.mul(frame,
+                                seg.sample_width,
+                                db_to_float(-attenuation))
+        
+        output.append(frame)
+    
+    return seg._spawn(data=b''.join(output))
+
+
+# Invert the phase of the signal.
+
+@register_pydub_effect
+
+def invert_phase(seg, channels=(1, 1)):
+    """
+    channels- specifies which channel (left or right) to reverse the phase of.
+    Note that mono AudioSegments will become stereo.
+    """
+    if channels == (1, 1):
+        inverted = audioop.mul(seg._data, seg.sample_width, -1.0)  
+        return seg._spawn(data=inverted)
+    
+    else:
+        if seg.channels == 2:
+            left, right = seg.split_to_mono()
+        else:
+            raise Exception("Can't implicitly convert an AudioSegment with " + str(seg.channels) + " channels to stereo.")
+            
+        if channels == (1, 0):    
+            left = left.invert_phase()
+        else:
+            right = right.invert_phase()
+        
+        return seg.from_mono_audiosegments(left, right)
+        
+
+
+# High and low pass filters based on implementation found on Stack Overflow:
+#   http://stackoverflow.com/questions/13882038/implementing-simple-high-and-low-pass-filters-in-c
+
+@register_pydub_effect
+def low_pass_filter(seg, cutoff):
+    """
+        cutoff - Frequency (in Hz) where higher frequency signal will begin to
+            be reduced by 6dB per octave (doubling in frequency) above this point
+    """
+    RC = 1.0 / (cutoff * 2 * math.pi)
+    dt = 1.0 / seg.frame_rate
+
+    alpha = dt / (RC + dt)
+    
+    original = seg.get_array_of_samples()
+    filteredArray = array.array(seg.array_type, original)
+    
+    frame_count = int(seg.frame_count())
+
+    last_val = [0] * seg.channels
+    for i in range(seg.channels):
+        last_val[i] = filteredArray[i] = original[i]
+
+    for i in range(1, frame_count):
+        for j in range(seg.channels):
+            offset = (i * seg.channels) + j
+            last_val[j] = last_val[j] + (alpha * (original[offset] - last_val[j]))
+            filteredArray[offset] = int(last_val[j])
+
+    return seg._spawn(data=filteredArray)
+
+
+@register_pydub_effect
+def high_pass_filter(seg, cutoff):
+    """
+        cutoff - Frequency (in Hz) where lower frequency signal will begin to
+            be reduced by 6dB per octave (doubling in frequency) below this point
+    """
+    RC = 1.0 / (cutoff * 2 * math.pi)
+    dt = 1.0 / seg.frame_rate
+
+    alpha = RC / (RC + dt)
+
+    minval, maxval = get_min_max_value(seg.sample_width * 8)
+    
+    original = seg.get_array_of_samples()
+    filteredArray = array.array(seg.array_type, original)
+    
+    frame_count = int(seg.frame_count())
+
+    last_val = [0] * seg.channels
+    for i in range(seg.channels):
+        last_val[i] = filteredArray[i] = original[i]
+
+    for i in range(1, frame_count):
+        for j in range(seg.channels):
+            offset = (i * seg.channels) + j
+            offset_minus_1 = ((i-1) * seg.channels) + j
+
+            last_val[j] = alpha * (last_val[j] + original[offset] - original[offset_minus_1])
+            filteredArray[offset] = int(min(max(last_val[j], minval), maxval))
+
+    return seg._spawn(data=filteredArray)
+    
+    
+@register_pydub_effect
+def pan(seg, pan_amount):
+    """
+    pan_amount should be between -1.0 (100% left) and +1.0 (100% right)
+    
+    When pan_amount == 0.0 the left/right balance is not changed.
+    
+    Panning does not alter the *perceived* loundness, but since loudness
+    is decreasing on one side, the other side needs to get louder to
+    compensate. When panned hard left, the left channel will be 3dB louder.
+    """
+    if not -1.0 <= pan_amount <= 1.0:
+        raise ValueError("pan_amount should be between -1.0 (100% left) and +1.0 (100% right)")
+    
+    max_boost_db = ratio_to_db(2.0)
+    boost_db = abs(pan_amount) * max_boost_db
+    
+    boost_factor = db_to_float(boost_db)
+    reduce_factor = db_to_float(max_boost_db) - boost_factor
+    
+    reduce_db = ratio_to_db(reduce_factor)
+    
+    # Cut boost in half (max boost== 3dB) - in reality 2 speakers
+    #   do not sum to a full 6 dB.
+    boost_db = boost_db / 2.0
+    
+    if pan_amount < 0:
+        return seg.apply_gain_stereo(boost_db, reduce_db)
+    else:
+        return seg.apply_gain_stereo(reduce_db, boost_db)
+        
+    
+@register_pydub_effect
+def apply_gain_stereo(seg, left_gain=0.0, right_gain=0.0):
+    """
+    left_gain - amount of gain to apply to the left channel (in dB)
+    right_gain - amount of gain to apply to the right channel (in dB)
+    
+    note: mono audio segments will be converted to stereo
+    """
+    if seg.channels == 1:
+        left = right = seg
+    elif seg.channels == 2:
+        left, right = seg.split_to_mono()
+    
+    l_mult_factor = db_to_float(left_gain)
+    r_mult_factor = db_to_float(right_gain)
+    
+    left_data = audioop.mul(left._data, left.sample_width, l_mult_factor)
+    left_data = audioop.tostereo(left_data, left.sample_width, 1, 0)
+    
+    right_data = audioop.mul(right._data, right.sample_width, r_mult_factor)
+    right_data = audioop.tostereo(right_data, right.sample_width, 0, 1)
+    
+    output = audioop.add(left_data, right_data, seg.sample_width)
+    
+    return seg._spawn(data=output,
+                overrides={'channels': 2,
+                           'frame_width': 2 * seg.sample_width})
@@ -0,0 +1,32 @@
+class PydubException(Exception):
+    """
+    Base class for any Pydub exception
+    """
+
+
+class TooManyMissingFrames(PydubException):
+    pass
+
+
+class InvalidDuration(PydubException):
+    pass
+
+
+class InvalidTag(PydubException):
+    pass
+
+
+class InvalidID3TagVersion(PydubException):
+    pass
+
+
+class CouldntDecodeError(PydubException):
+    pass
+
+
+class CouldntEncodeError(PydubException):
+    pass
+
+
+class MissingAudioParameter(PydubException):
+    pass
@@ -0,0 +1,142 @@
+"""
+Each generator will return float samples from -1.0 to 1.0, which can be 
+converted to actual audio with 8, 16, 24, or 32 bit depth using the
+SiganlGenerator.to_audio_segment() method (on any of it's subclasses).
+
+See Wikipedia's "waveform" page for info on some of the generators included 
+here: http://en.wikipedia.org/wiki/Waveform
+"""
+
+import math
+import array
+import itertools
+import random
+from .audio_segment import AudioSegment
+from .utils import (
+    db_to_float,
+    get_frame_width,
+    get_array_type,
+    get_min_max_value
+)
+
+
+
+class SignalGenerator(object):
+    def __init__(self, sample_rate=44100, bit_depth=16):
+        self.sample_rate = sample_rate
+        self.bit_depth = bit_depth
+
+    def to_audio_segment(self, duration=1000.0, volume=0.0):
+        """
+        Duration in milliseconds
+            (default: 1 second)
+        Volume in DB relative to maximum amplitude
+            (default 0.0 dBFS, which is the maximum value)
+        """
+        minval, maxval = get_min_max_value(self.bit_depth)
+        sample_width = get_frame_width(self.bit_depth)
+        array_type = get_array_type(self.bit_depth)
+
+        gain = db_to_float(volume)
+        sample_count = int(self.sample_rate * (duration / 1000.0))
+
+        sample_data = (int(val * maxval * gain) for val in self.generate())
+        sample_data = itertools.islice(sample_data, 0, sample_count)
+
+        data = array.array(array_type, sample_data)
+        
+        try:
+            data = data.tobytes()
+        except:
+            data = data.tostring()
+
+        return AudioSegment(data=data, metadata={
+            "channels": 1,
+            "sample_width": sample_width,
+            "frame_rate": self.sample_rate,
+            "frame_width": sample_width,
+        })
+
+    def generate(self):
+        raise NotImplementedError("SignalGenerator subclasses must implement the generate() method, and *should not* call the superclass implementation.")
+
+
+
+class Sine(SignalGenerator):
+    def __init__(self, freq, **kwargs):
+        super(Sine, self).__init__(**kwargs)
+        self.freq = freq
+
+    def generate(self):
+        sine_of = (self.freq * 2 * math.pi) / self.sample_rate
+        sample_n = 0
+        while True:
+            yield math.sin(sine_of * sample_n)
+            sample_n += 1
+
+
+
+class Pulse(SignalGenerator):
+    def __init__(self, freq, duty_cycle=0.5, **kwargs):
+        super(Pulse, self).__init__(**kwargs)
+        self.freq = freq
+        self.duty_cycle = duty_cycle
+
+    def generate(self):
+        sample_n = 0
+
+        # in samples
+        cycle_length = self.sample_rate / float(self.freq)
+        pulse_length = cycle_length * self.duty_cycle
+
+        while True:
+            if (sample_n % cycle_length) < pulse_length:
+                yield 1.0
+            else:
+                yield -1.0
+            sample_n += 1
+
+
+
+class Square(Pulse):
+    def __init__(self, freq, **kwargs):
+        kwargs['duty_cycle'] = 0.5
+        super(Square, self).__init__(freq, **kwargs)
+
+
+
+class Sawtooth(SignalGenerator):
+    def __init__(self, freq, duty_cycle=1.0, **kwargs):
+        super(Sawtooth, self).__init__(**kwargs)
+        self.freq = freq
+        self.duty_cycle = duty_cycle
+
+    def generate(self):
+        sample_n = 0
+
+        # in samples
+        cycle_length = self.sample_rate / float(self.freq)
+        midpoint = cycle_length * self.duty_cycle
+        ascend_length = midpoint
+        descend_length = cycle_length - ascend_length
+
+        while True:
+            cycle_position = sample_n % cycle_length
+            if cycle_position < midpoint:
+                yield (2 * cycle_position / ascend_length) - 1.0
+            else:
+                yield 1.0 - (2 * (cycle_position - midpoint) / descend_length)
+            sample_n += 1
+
+
+
+class Triangle(Sawtooth):
+    def __init__(self, freq, **kwargs):
+        kwargs['duty_cycle'] = 0.5
+        super(Triangle, self).__init__(freq, **kwargs)
+
+
+class WhiteNoise(SignalGenerator):
+    def generate(self):
+        while True:
+            yield (random.random() * 2) - 1.0
@@ -0,0 +1,14 @@
+"""
+
+"""
+import logging
+
+converter_logger = logging.getLogger("swingmusic.pydub.converter")
+
+def log_conversion(conversion_command):
+    converter_logger.debug("subprocess.call(%s)", repr(conversion_command))
+
+def log_subprocess_output(output):
+    if output:
+        for line in output.rstrip().splitlines():
+            converter_logger.debug('subprocess output: %s', line.rstrip())
@@ -0,0 +1,71 @@
+"""
+Support for playing AudioSegments. Pyaudio will be used if it's installed,
+otherwise will fallback to ffplay. Pyaudio is a *much* nicer solution, but
+is tricky to install. See my notes on installing pyaudio in a virtualenv (on
+OSX 10.10): https://gist.github.com/jiaaro/9767512210a1d80a8a0d
+"""
+
+import subprocess
+from tempfile import NamedTemporaryFile
+from .utils import get_player_name, make_chunks
+
+def _play_with_ffplay(seg):
+    PLAYER = get_player_name()
+    with NamedTemporaryFile("w+b", suffix=".wav") as f:
+        seg.export(f.name, "wav")
+        subprocess.call([PLAYER, "-nodisp", "-autoexit", "-hide_banner", f.name])
+
+
+def _play_with_pyaudio(seg):
+    import pyaudio
+
+    p = pyaudio.PyAudio()
+    stream = p.open(format=p.get_format_from_width(seg.sample_width),
+                    channels=seg.channels,
+                    rate=seg.frame_rate,
+                    output=True)
+
+    # Just in case there were any exceptions/interrupts, we release the resource
+    # So as not to raise OSError: Device Unavailable should play() be used again
+    try:
+        # break audio into half-second chunks (to allows keyboard interrupts)
+        for chunk in make_chunks(seg, 500):
+            stream.write(chunk._data)
+    finally:
+        stream.stop_stream()
+        stream.close()
+
+        p.terminate()
+
+
+def _play_with_simpleaudio(seg):
+    import simpleaudio
+    return simpleaudio.play_buffer(
+        seg.raw_data,
+        num_channels=seg.channels,
+        bytes_per_sample=seg.sample_width,
+        sample_rate=seg.frame_rate
+    )
+
+
+def play(audio_segment):
+    try:
+        playback = _play_with_simpleaudio(audio_segment)
+        try:
+            playback.wait_done()
+        except KeyboardInterrupt:
+            playback.stop()
+    except ImportError:
+        pass
+    else:
+        return
+
+    try:
+        _play_with_pyaudio(audio_segment)
+        return
+    except ImportError:
+        pass
+    else:
+        return
+
+    _play_with_ffplay(audio_segment)
@@ -0,0 +1,553 @@
+try:
+    from __builtin__ import max as builtin_max
+    from __builtin__ import min as builtin_min
+except ImportError:
+    from builtins import max as builtin_max
+    from builtins import min as builtin_min
+import math
+import struct
+try:
+    from fractions import gcd
+except ImportError:  # Python 3.9+
+    from math import gcd
+from ctypes import create_string_buffer
+
+
+class error(Exception):
+    pass
+
+
+def _check_size(size):
+    if size != 1 and size != 2 and size != 4:
+        raise error("Size should be 1, 2 or 4")
+
+
+def _check_params(length, size):
+    _check_size(size)
+    if length % size != 0:
+        raise error("not a whole number of frames")
+
+
+def _sample_count(cp, size):
+    return len(cp) / size
+
+
+def _get_samples(cp, size, signed=True):
+    for i in range(_sample_count(cp, size)):
+        yield _get_sample(cp, size, i, signed)
+
+
+def _struct_format(size, signed):
+    if size == 1:
+        return "b" if signed else "B"
+    elif size == 2:
+        return "h" if signed else "H"
+    elif size == 4:
+        return "i" if signed else "I"
+
+
+def _get_sample(cp, size, i, signed=True):
+    fmt = _struct_format(size, signed)
+    start = i * size
+    end = start + size
+    return struct.unpack_from(fmt, buffer(cp)[start:end])[0]
+
+
+def _put_sample(cp, size, i, val, signed=True):
+    fmt = _struct_format(size, signed)
+    struct.pack_into(fmt, cp, i * size, val)
+
+
+def _get_maxval(size, signed=True):
+    if signed and size == 1:
+        return 0x7f
+    elif size == 1:
+        return 0xff
+    elif signed and size == 2:
+        return 0x7fff
+    elif size == 2:
+        return 0xffff
+    elif signed and size == 4:
+        return 0x7fffffff
+    elif size == 4:
+        return 0xffffffff
+
+
+def _get_minval(size, signed=True):
+    if not signed:
+        return 0
+    elif size == 1:
+        return -0x80
+    elif size == 2:
+        return -0x8000
+    elif size == 4:
+        return -0x80000000
+
+
+def _get_clipfn(size, signed=True):
+    maxval = _get_maxval(size, signed)
+    minval = _get_minval(size, signed)
+    return lambda val: builtin_max(min(val, maxval), minval)
+
+
+def _overflow(val, size, signed=True):
+    minval = _get_minval(size, signed)
+    maxval = _get_maxval(size, signed)
+    if minval <= val <= maxval:
+        return val
+
+    bits = size * 8
+    if signed:
+        offset = 2**(bits-1)
+        return ((val + offset) % (2**bits)) - offset
+    else:
+        return val % (2**bits)
+
+
+def getsample(cp, size, i):
+    _check_params(len(cp), size)
+    if not (0 <= i < len(cp) / size):
+        raise error("Index out of range")
+    return _get_sample(cp, size, i)
+
+
+def max(cp, size):
+    _check_params(len(cp), size)
+
+    if len(cp) == 0:
+        return 0
+
+    return builtin_max(abs(sample) for sample in _get_samples(cp, size))
+
+
+def minmax(cp, size):
+    _check_params(len(cp), size)
+
+    max_sample, min_sample = 0, 0
+    for sample in _get_samples(cp, size):
+        max_sample = builtin_max(sample, max_sample)
+        min_sample = builtin_min(sample, min_sample)
+
+    return min_sample, max_sample
+
+
+def avg(cp, size):
+    _check_params(len(cp), size)
+    sample_count = _sample_count(cp, size)
+    if sample_count == 0:
+        return 0
+    return sum(_get_samples(cp, size)) / sample_count
+
+
+def rms(cp, size):
+    _check_params(len(cp), size)
+
+    sample_count = _sample_count(cp, size)
+    if sample_count == 0:
+        return 0
+
+    sum_squares = sum(sample**2 for sample in _get_samples(cp, size))
+    return int(math.sqrt(sum_squares / sample_count))
+
+
+def _sum2(cp1, cp2, length):
+    size = 2
+    total = 0
+    for i in range(length):
+        total += getsample(cp1, size, i) * getsample(cp2, size, i)
+    return total
+
+
+def findfit(cp1, cp2):
+    size = 2
+
+    if len(cp1) % 2 != 0 or len(cp2) % 2 != 0:
+        raise error("Strings should be even-sized")
+
+    if len(cp1) < len(cp2):
+        raise error("First sample should be longer")
+
+    len1 = _sample_count(cp1, size)
+    len2 = _sample_count(cp2, size)
+
+    sum_ri_2 = _sum2(cp2, cp2, len2)
+    sum_aij_2 = _sum2(cp1, cp1, len2)
+    sum_aij_ri = _sum2(cp1, cp2, len2)
+
+    result = (sum_ri_2 * sum_aij_2 - sum_aij_ri * sum_aij_ri) / sum_aij_2
+
+    best_result = result
+    best_i = 0
+
+    for i in range(1, len1 - len2 + 1):
+        aj_m1 = _get_sample(cp1, size, i - 1)
+        aj_lm1 = _get_sample(cp1, size, i + len2 - 1)
+
+        sum_aij_2 += aj_lm1**2 - aj_m1**2
+        sum_aij_ri = _sum2(buffer(cp1)[i*size:], cp2, len2)
+
+        result = (sum_ri_2 * sum_aij_2 - sum_aij_ri * sum_aij_ri) / sum_aij_2
+
+        if result < best_result:
+            best_result = result
+            best_i = i
+
+    factor = _sum2(buffer(cp1)[best_i*size:], cp2, len2) / sum_ri_2
+
+    return best_i, factor
+
+
+def findfactor(cp1, cp2):
+    size = 2
+
+    if len(cp1) % 2 != 0:
+        raise error("Strings should be even-sized")
+
+    if len(cp1) != len(cp2):
+        raise error("Samples should be same size")
+
+    sample_count = _sample_count(cp1, size)
+
+    sum_ri_2 = _sum2(cp2, cp2, sample_count)
+    sum_aij_ri = _sum2(cp1, cp2, sample_count)
+
+    return sum_aij_ri / sum_ri_2
+
+
+def findmax(cp, len2):
+    size = 2
+    sample_count = _sample_count(cp, size)
+
+    if len(cp) % 2 != 0:
+        raise error("Strings should be even-sized")
+
+    if len2 < 0 or sample_count < len2:
+        raise error("Input sample should be longer")
+
+    if sample_count == 0:
+        return 0
+
+    result = _sum2(cp, cp, len2)
+    best_result = result
+    best_i = 0
+
+    for i in range(1, sample_count - len2 + 1):
+        sample_leaving_window = getsample(cp, size, i - 1)
+        sample_entering_window = getsample(cp, size, i + len2 - 1)
+
+        result -= sample_leaving_window**2
+        result += sample_entering_window**2
+
+        if result > best_result:
+            best_result = result
+            best_i = i
+
+    return best_i
+
+
+def avgpp(cp, size):
+    _check_params(len(cp), size)
+    sample_count = _sample_count(cp, size)
+
+    prevextremevalid = False
+    prevextreme = None
+    avg = 0
+    nextreme = 0
+
+    prevval = getsample(cp, size, 0)
+    val = getsample(cp, size, 1)
+
+    prevdiff = val - prevval
+
+    for i in range(1, sample_count):
+        val = getsample(cp, size, i)
+        diff = val - prevval
+
+        if diff * prevdiff < 0:
+            if prevextremevalid:
+                avg += abs(prevval - prevextreme)
+                nextreme += 1
+
+            prevextremevalid = True
+            prevextreme = prevval
+
+        prevval = val
+        if diff != 0:
+            prevdiff = diff
+
+    if nextreme == 0:
+        return 0
+
+    return avg / nextreme
+
+
+def maxpp(cp, size):
+    _check_params(len(cp), size)
+    sample_count = _sample_count(cp, size)
+
+    prevextremevalid = False
+    prevextreme = None
+    max = 0
+
+    prevval = getsample(cp, size, 0)
+    val = getsample(cp, size, 1)
+
+    prevdiff = val - prevval
+
+    for i in range(1, sample_count):
+        val = getsample(cp, size, i)
+        diff = val - prevval
+
+        if diff * prevdiff < 0:
+            if prevextremevalid:
+                extremediff = abs(prevval - prevextreme)
+                if extremediff > max:
+                    max = extremediff
+            prevextremevalid = True
+            prevextreme = prevval
+
+        prevval = val
+        if diff != 0:
+            prevdiff = diff
+
+    return max
+
+
+def cross(cp, size):
+    _check_params(len(cp), size)
+
+    crossings = 0
+    last_sample = 0
+    for sample in _get_samples(cp, size):
+        if sample <= 0 < last_sample or sample >= 0 > last_sample:
+            crossings += 1
+        last_sample = sample
+
+    return crossings
+
+
+def mul(cp, size, factor):
+    _check_params(len(cp), size)
+    clip = _get_clipfn(size)
+
+    result = create_string_buffer(len(cp))
+
+    for i, sample in enumerate(_get_samples(cp, size)):
+        sample = clip(int(sample * factor))
+        _put_sample(result, size, i, sample)
+
+    return result.raw
+
+
+def tomono(cp, size, fac1, fac2):
+    _check_params(len(cp), size)
+    clip = _get_clipfn(size)
+
+    sample_count = _sample_count(cp, size)
+
+    result = create_string_buffer(len(cp) / 2)
+
+    for i in range(0, sample_count, 2):
+        l_sample = getsample(cp, size, i)
+        r_sample = getsample(cp, size, i + 1)
+
+        sample = (l_sample * fac1) + (r_sample * fac2)
+        sample = clip(sample)
+
+        _put_sample(result, size, i / 2, sample)
+
+    return result.raw
+
+
+def tostereo(cp, size, fac1, fac2):
+    _check_params(len(cp), size)
+
+    sample_count = _sample_count(cp, size)
+
+    result = create_string_buffer(len(cp) * 2)
+    clip = _get_clipfn(size)
+
+    for i in range(sample_count):
+        sample = _get_sample(cp, size, i)
+
+        l_sample = clip(sample * fac1)
+        r_sample = clip(sample * fac2)
+
+        _put_sample(result, size, i * 2, l_sample)
+        _put_sample(result, size, i * 2 + 1, r_sample)
+
+    return result.raw
+
+
+def add(cp1, cp2, size):
+    _check_params(len(cp1), size)
+
+    if len(cp1) != len(cp2):
+        raise error("Lengths should be the same")
+
+    clip = _get_clipfn(size)
+    sample_count = _sample_count(cp1, size)
+    result = create_string_buffer(len(cp1))
+
+    for i in range(sample_count):
+        sample1 = getsample(cp1, size, i)
+        sample2 = getsample(cp2, size, i)
+
+        sample = clip(sample1 + sample2)
+
+        _put_sample(result, size, i, sample)
+
+    return result.raw
+
+
+def bias(cp, size, bias):
+    _check_params(len(cp), size)
+
+    result = create_string_buffer(len(cp))
+
+    for i, sample in enumerate(_get_samples(cp, size)):
+        sample = _overflow(sample + bias, size)
+        _put_sample(result, size, i, sample)
+
+    return result.raw
+
+
+def reverse(cp, size):
+    _check_params(len(cp), size)
+    sample_count = _sample_count(cp, size)
+
+    result = create_string_buffer(len(cp))
+    for i, sample in enumerate(_get_samples(cp, size)):
+        _put_sample(result, size, sample_count - i - 1, sample)
+
+    return result.raw
+
+
+def lin2lin(cp, size, size2):
+    _check_params(len(cp), size)
+    _check_size(size2)
+
+    if size == size2:
+        return cp
+
+    new_len = (len(cp) / size) * size2
+
+    result = create_string_buffer(new_len)
+
+    for i in range(_sample_count(cp, size)):
+        sample = _get_sample(cp, size, i)
+        if size < size2:
+            sample = sample << (4 * size2 / size)
+        elif size > size2:
+            sample = sample >> (4 * size / size2)
+
+        sample = _overflow(sample, size2)
+
+        _put_sample(result, size2, i, sample)
+
+    return result.raw
+
+
+def ratecv(cp, size, nchannels, inrate, outrate, state, weightA=1, weightB=0):
+    _check_params(len(cp), size)
+    if nchannels < 1:
+        raise error("# of channels should be >= 1")
+
+    bytes_per_frame = size * nchannels
+    frame_count = len(cp) / bytes_per_frame
+
+    if bytes_per_frame / nchannels != size:
+        raise OverflowError("width * nchannels too big for a C int")
+
+    if weightA < 1 or weightB < 0:
+        raise error("weightA should be >= 1, weightB should be >= 0")
+
+    if len(cp) % bytes_per_frame != 0:
+        raise error("not a whole number of frames")
+
+    if inrate <= 0 or outrate <= 0:
+        raise error("sampling rate not > 0")
+
+    d = gcd(inrate, outrate)
+    inrate /= d
+    outrate /= d
+
+    prev_i = [0] * nchannels
+    cur_i = [0] * nchannels
+
+    if state is None:
+        d = -outrate
+    else:
+        d, samps = state
+
+        if len(samps) != nchannels:
+            raise error("illegal state argument")
+
+        prev_i, cur_i = zip(*samps)
+        prev_i, cur_i = list(prev_i), list(cur_i)
+
+    q = frame_count / inrate
+    ceiling = (q + 1) * outrate
+    nbytes = ceiling * bytes_per_frame
+
+    result = create_string_buffer(nbytes)
+
+    samples = _get_samples(cp, size)
+    out_i = 0
+    while True:
+        while d < 0:
+            if frame_count == 0:
+                samps = zip(prev_i, cur_i)
+                retval = result.raw
+
+                # slice off extra bytes
+                trim_index = (out_i * bytes_per_frame) - len(retval)
+                retval = buffer(retval)[:trim_index]
+
+                return (retval, (d, tuple(samps)))
+
+            for chan in range(nchannels):
+                prev_i[chan] = cur_i[chan]
+                cur_i[chan] = samples.next()
+
+                cur_i[chan] = (
+                    (weightA * cur_i[chan] + weightB * prev_i[chan])
+                    / (weightA + weightB)
+                )
+
+            frame_count -= 1
+            d += outrate
+
+        while d >= 0:
+            for chan in range(nchannels):
+                cur_o = (
+                    (prev_i[chan] * d + cur_i[chan] * (outrate - d))
+                    / outrate
+                )
+                _put_sample(result, size, out_i, _overflow(cur_o, size))
+                out_i += 1
+            d -= inrate
+
+
+def lin2ulaw(cp, size):
+    raise NotImplementedError()
+
+
+def ulaw2lin(cp, size):
+    raise NotImplementedError()
+
+
+def lin2alaw(cp, size):
+    raise NotImplementedError()
+
+
+def alaw2lin(cp, size):
+    raise NotImplementedError()
+
+
+def lin2adpcm(cp, size, state):
+    raise NotImplementedError()
+
+
+def adpcm2lin(cp, size, state):
+    raise NotImplementedError()
@@ -0,0 +1,175 @@
+"""
+This module provides scipy versions of high_pass_filter, and low_pass_filter
+as well as an additional band_pass_filter.
+
+Of course, you will need to install scipy for these to work.
+
+When this module is imported the high and low pass filters from this module
+will be used when calling audio_segment.high_pass_filter() and
+audio_segment.high_pass_filter() instead of the slower, less powerful versions
+provided by pydub.effects.
+"""
+from scipy.signal import butter, sosfilt
+from .utils import (register_pydub_effect,stereo_to_ms,ms_to_stereo)
+
+
+def _mk_butter_filter(freq, type, order):
+    """
+    Args:
+        freq: The cutoff frequency for highpass and lowpass filters. For
+            band filters, a list of [low_cutoff, high_cutoff]
+        type: "lowpass", "highpass", or "band"
+        order: nth order butterworth filter (default: 5th order). The
+            attenuation is -6dB/octave beyond the cutoff frequency (for 1st
+            order). A Higher order filter will have more attenuation, each level
+            adding an additional -6dB (so a 3rd order butterworth filter would
+            be -18dB/octave).
+
+    Returns:
+        function which can filter a mono audio segment
+
+    """
+    def filter_fn(seg):
+        assert seg.channels == 1
+
+        nyq = 0.5 * seg.frame_rate
+        try:
+            freqs = [f / nyq for f in freq]
+        except TypeError:
+            freqs = freq / nyq
+
+        sos = butter(order, freqs, btype=type, output='sos')
+        y = sosfilt(sos, seg.get_array_of_samples())
+
+        return seg._spawn(y.astype(seg.array_type))
+
+    return filter_fn
+
+
+@register_pydub_effect
+def band_pass_filter(seg, low_cutoff_freq, high_cutoff_freq, order=5):
+    filter_fn = _mk_butter_filter([low_cutoff_freq, high_cutoff_freq], 'band', order=order)
+    return seg.apply_mono_filter_to_each_channel(filter_fn)
+
+
+@register_pydub_effect
+def high_pass_filter(seg, cutoff_freq, order=5):
+    filter_fn = _mk_butter_filter(cutoff_freq, 'highpass', order=order)
+    return seg.apply_mono_filter_to_each_channel(filter_fn)
+
+
+@register_pydub_effect
+def low_pass_filter(seg, cutoff_freq, order=5):
+    filter_fn = _mk_butter_filter(cutoff_freq, 'lowpass', order=order)
+    return seg.apply_mono_filter_to_each_channel(filter_fn)
+
+
+@register_pydub_effect
+def _eq(seg, focus_freq, bandwidth=100, mode="peak", gain_dB=0, order=2):
+    """
+    Args:
+        focus_freq - middle frequency or known frequency of band (in Hz)
+        bandwidth - range of the equalizer band
+        mode - Mode of Equalization(Peak/Notch(Bell Curve),High Shelf, Low Shelf)
+        order - Rolloff factor(1 - 6dB/Octave 2 - 12dB/Octave)
+    
+    Returns:
+        Equalized/Filtered AudioSegment
+    """
+    filt_mode = ["peak", "low_shelf", "high_shelf"]
+    if mode not in filt_mode:
+        raise ValueError("Incorrect Mode Selection")
+        
+    if gain_dB >= 0:
+        if mode == "peak":
+            sec = band_pass_filter(seg, focus_freq - bandwidth/2, focus_freq + bandwidth/2, order = order)
+            seg = seg.overlay(sec - (3 - gain_dB))
+            return seg
+        
+        if mode == "low_shelf":
+            sec = low_pass_filter(seg, focus_freq, order=order)
+            seg = seg.overlay(sec - (3 - gain_dB))
+            return seg
+        
+        if mode == "high_shelf":
+            sec = high_pass_filter(seg, focus_freq, order=order)
+            seg = seg.overlay(sec - (3 - gain_dB))
+            return seg
+        
+    if gain_dB < 0:
+        if mode == "peak":
+            sec = high_pass_filter(seg, focus_freq - bandwidth/2, order=order)
+            seg = seg.overlay(sec - (3 + gain_dB)) + gain_dB
+            sec = low_pass_filter(seg, focus_freq + bandwidth/2, order=order)
+            seg = seg.overlay(sec - (3 + gain_dB)) + gain_dB
+            return seg
+        
+        if mode == "low_shelf":
+            sec = high_pass_filter(seg, focus_freq, order=order)
+            seg = seg.overlay(sec - (3 + gain_dB)) + gain_dB
+            return seg
+        
+        if mode=="high_shelf":
+            sec=low_pass_filter(seg, focus_freq, order=order)
+            seg=seg.overlay(sec - (3 + gain_dB)) +gain_dB
+            return seg
+        
+
+@register_pydub_effect
+def eq(seg, focus_freq, bandwidth=100, channel_mode="L+R", filter_mode="peak", gain_dB=0, order=2):
+    """
+    Args:
+        focus_freq - middle frequency or known frequency of band (in Hz)
+        bandwidth - range of the equalizer band
+        channel_mode - Select Channels to be affected by the filter.
+            L+R - Standard Stereo Filter
+            L - Only Left Channel is Filtered
+            R - Only Right Channel is Filtered
+            M+S - Blumlien Stereo Filter(Mid-Side)
+            M - Only Mid Channel is Filtered
+            S - Only Side Channel is Filtered
+            Mono Audio Segments are completely filtered.
+        filter_mode - Mode of Equalization(Peak/Notch(Bell Curve),High Shelf, Low Shelf)
+        order - Rolloff factor(1 - 6dB/Octave 2 - 12dB/Octave)
+    
+    Returns:
+        Equalized/Filtered AudioSegment
+    """
+    channel_modes = ["L+R", "M+S", "L", "R", "M", "S"]
+    if channel_mode not in channel_modes:
+        raise ValueError("Incorrect Channel Mode Selection")
+        
+    if seg.channels == 1:
+        return _eq(seg, focus_freq, bandwidth, filter_mode, gain_dB, order)
+        
+    if channel_mode == "L+R":
+        return _eq(seg, focus_freq, bandwidth, filter_mode, gain_dB, order)
+        
+    if channel_mode == "L":
+        seg = seg.split_to_mono()
+        seg = [_eq(seg[0], focus_freq, bandwidth, filter_mode, gain_dB, order), seg[1]]
+        return AudioSegment.from_mono_audio_segements(seg[0], seg[1])
+        
+    if channel_mode == "R":
+        seg = seg.split_to_mono()
+        seg = [seg[0], _eq(seg[1], focus_freq, bandwidth, filter_mode, gain_dB, order)]
+        return AudioSegment.from_mono_audio_segements(seg[0], seg[1])
+        
+    if channel_mode == "M+S":
+        seg = stereo_to_ms(seg)
+        seg = _eq(seg, focus_freq, bandwidth, filter_mode, gain_dB, order)
+        return ms_to_stereo(seg)
+        
+    if channel_mode == "M":
+        seg = stereo_to_ms(seg).split_to_mono()
+        seg = [_eq(seg[0], focus_freq, bandwidth, filter_mode, gain_dB, order), seg[1]]
+        seg = AudioSegment.from_mono_audio_segements(seg[0], seg[1])
+        return ms_to_stereo(seg)
+        
+    if channel_mode == "S":
+        seg = stereo_to_ms(seg).split_to_mono()
+        seg = [seg[0], _eq(seg[1], focus_freq, bandwidth, filter_mode, gain_dB, order)]
+        seg = AudioSegment.from_mono_audio_segements(seg[0], seg[1])
+        return ms_to_stereo(seg)
+
+
@@ -0,0 +1,182 @@
+"""
+Various functions for finding/manipulating silence in AudioSegments
+"""
+import itertools
+
+from .utils import db_to_float
+
+
+def detect_silence(audio_segment, min_silence_len=1000, silence_thresh=-16, seek_step=1):
+    """
+    Returns a list of all silent sections [start, end] in milliseconds of audio_segment.
+    Inverse of detect_nonsilent()
+
+    audio_segment - the segment to find silence in
+    min_silence_len - the minimum length for any silent section
+    silence_thresh - the upper bound for how quiet is silent in dFBS
+    seek_step - step size for interating over the segment in ms
+    """
+    seg_len = len(audio_segment)
+
+    # you can't have a silent portion of a sound that is longer than the sound
+    if seg_len < min_silence_len:
+        return []
+
+    # convert silence threshold to a float value (so we can compare it to rms)
+    silence_thresh = db_to_float(silence_thresh) * audio_segment.max_possible_amplitude
+
+    # find silence and add start and end indicies to the to_cut list
+    silence_starts = []
+
+    # check successive (1 sec by default) chunk of sound for silence
+    # try a chunk at every "seek step" (or every chunk for a seek step == 1)
+    last_slice_start = seg_len - min_silence_len
+    slice_starts = range(0, last_slice_start + 1, seek_step)
+
+    # guarantee last_slice_start is included in the range
+    # to make sure the last portion of the audio is searched
+    if last_slice_start % seek_step:
+        slice_starts = itertools.chain(slice_starts, [last_slice_start])
+
+    for i in slice_starts:
+        audio_slice = audio_segment[i:i + min_silence_len]
+        if audio_slice.rms <= silence_thresh:
+            silence_starts.append(i)
+
+    # short circuit when there is no silence
+    if not silence_starts:
+        return []
+
+    # combine the silence we detected into ranges (start ms - end ms)
+    silent_ranges = []
+
+    prev_i = silence_starts.pop(0)
+    current_range_start = prev_i
+
+    for silence_start_i in silence_starts:
+        continuous = (silence_start_i == prev_i + seek_step)
+
+        # sometimes two small blips are enough for one particular slice to be
+        # non-silent, despite the silence all running together. Just combine
+        # the two overlapping silent ranges.
+        silence_has_gap = silence_start_i > (prev_i + min_silence_len)
+
+        if not continuous and silence_has_gap:
+            silent_ranges.append([current_range_start,
+                                  prev_i + min_silence_len])
+            current_range_start = silence_start_i
+        prev_i = silence_start_i
+
+    silent_ranges.append([current_range_start,
+                          prev_i + min_silence_len])
+
+    return silent_ranges
+
+
+def detect_nonsilent(audio_segment, min_silence_len=1000, silence_thresh=-16, seek_step=1):
+    """
+    Returns a list of all nonsilent sections [start, end] in milliseconds of audio_segment.
+    Inverse of detect_silent()
+
+    audio_segment - the segment to find silence in
+    min_silence_len - the minimum length for any silent section
+    silence_thresh - the upper bound for how quiet is silent in dFBS
+    seek_step - step size for interating over the segment in ms
+    """
+    silent_ranges = detect_silence(audio_segment, min_silence_len, silence_thresh, seek_step)
+    len_seg = len(audio_segment)
+
+    # if there is no silence, the whole thing is nonsilent
+    if not silent_ranges:
+        return [[0, len_seg]]
+
+    # short circuit when the whole audio segment is silent
+    if silent_ranges[0][0] == 0 and silent_ranges[0][1] == len_seg:
+        return []
+
+    prev_end_i = 0
+    nonsilent_ranges = []
+    for start_i, end_i in silent_ranges:
+        nonsilent_ranges.append([prev_end_i, start_i])
+        prev_end_i = end_i
+
+    if end_i != len_seg:
+        nonsilent_ranges.append([prev_end_i, len_seg])
+
+    if nonsilent_ranges[0] == [0, 0]:
+        nonsilent_ranges.pop(0)
+
+    return nonsilent_ranges
+
+
+def split_on_silence(audio_segment, min_silence_len=1000, silence_thresh=-16, keep_silence=100,
+                     seek_step=1):
+    """
+    Returns list of audio segments from splitting audio_segment on silent sections
+
+    audio_segment - original pydub.AudioSegment() object
+
+    min_silence_len - (in ms) minimum length of a silence to be used for
+        a split. default: 1000ms
+
+    silence_thresh - (in dBFS) anything quieter than this will be
+        considered silence. default: -16dBFS
+
+    keep_silence - (in ms or True/False) leave some silence at the beginning
+        and end of the chunks. Keeps the sound from sounding like it
+        is abruptly cut off.
+        When the length of the silence is less than the keep_silence duration
+        it is split evenly between the preceding and following non-silent
+        segments.
+        If True is specified, all the silence is kept, if False none is kept.
+        default: 100ms
+
+    seek_step - step size for interating over the segment in ms
+    """
+
+    # from the itertools documentation
+    def pairwise(iterable):
+        "s -> (s0,s1), (s1,s2), (s2, s3), ..."
+        a, b = itertools.tee(iterable)
+        next(b, None)
+        return zip(a, b)
+
+    if isinstance(keep_silence, bool):
+        keep_silence = len(audio_segment) if keep_silence else 0
+
+    output_ranges = [
+        [ start - keep_silence, end + keep_silence ]
+        for (start,end)
+            in detect_nonsilent(audio_segment, min_silence_len, silence_thresh, seek_step)
+    ]
+
+    for range_i, range_ii in pairwise(output_ranges):
+        last_end = range_i[1]
+        next_start = range_ii[0]
+        if next_start < last_end:
+            range_i[1] = (last_end+next_start)//2
+            range_ii[0] = range_i[1]
+
+    return [
+        audio_segment[ max(start,0) : min(end,len(audio_segment)) ]
+        for start,end in output_ranges
+    ]
+
+
+def detect_leading_silence(sound, silence_threshold=-50.0, chunk_size=10):
+    """
+    Returns the millisecond/index that the leading silence ends.
+
+    audio_segment - the segment to find silence in
+    silence_threshold - the upper bound for how quiet is silent in dFBS
+    chunk_size - chunk size for interating over the segment in ms
+    """
+    trim_ms = 0 # ms
+    assert chunk_size > 0 # to avoid infinite loop
+    while sound[trim_ms:trim_ms+chunk_size].dBFS < silence_threshold and trim_ms < len(sound):
+        trim_ms += chunk_size
+
+    # if there is no end it should return the length of the segment
+    return min(trim_ms, len(sound))
+
+
@@ -0,0 +1,440 @@
+from __future__ import division
+from io import BufferedReader
+
+import json
+import os
+import re
+import sys
+from subprocess import Popen, PIPE
+from math import log, ceil
+from tempfile import TemporaryFile
+from warnings import warn
+from functools import wraps
+
+try:
+    import audioop
+except ImportError:
+    import pyaudioop as audioop
+
+if sys.version_info >= (3, 0):
+    basestring = str
+
+FRAME_WIDTHS = {
+    8: 1,
+    16: 2,
+    32: 4,
+}
+ARRAY_TYPES = {
+    8: "b",
+    16: "h",
+    32: "i",
+}
+ARRAY_RANGES = {
+    8: (-0x80, 0x7f),
+    16: (-0x8000, 0x7fff),
+    32: (-0x80000000, 0x7fffffff),
+}
+
+
+def get_frame_width(bit_depth):
+    return FRAME_WIDTHS[bit_depth]
+
+
+def get_array_type(bit_depth, signed=True):
+    t = ARRAY_TYPES[bit_depth]
+    if not signed:
+        t = t.upper()
+    return t
+
+
+def get_min_max_value(bit_depth):
+    return ARRAY_RANGES[bit_depth]
+
+
+def _fd_or_path_or_tempfile(fd, mode='w+b', tempfile=True):
+    close_fd = False
+    if fd is None and tempfile:
+        fd = TemporaryFile(mode=mode)
+        close_fd = True
+
+    if isinstance(fd, basestring):
+        fd = open(fd, mode=mode)
+        close_fd = True
+
+    if isinstance(fd, BufferedReader):
+        close_fd = True
+
+    try:
+        if isinstance(fd, os.PathLike):
+            fd = open(fd, mode=mode)
+            close_fd = True
+    except AttributeError:
+        # module os has no attribute PathLike, so we're on python < 3.6.
+        # The protocol we're trying to support doesn't exist, so just pass.
+        pass
+
+    return fd, close_fd
+
+
+def db_to_float(db, using_amplitude=True):
+    """
+    Converts the input db to a float, which represents the equivalent
+    ratio in power.
+    """
+    db = float(db)
+    if using_amplitude:
+        return 10 ** (db / 20)
+    else:  # using power
+        return 10 ** (db / 10)
+
+
+def ratio_to_db(ratio, val2=None, using_amplitude=True):
+    """
+    Converts the input float to db, which represents the equivalent
+    to the ratio in power represented by the multiplier passed in.
+    """
+    ratio = float(ratio)
+
+    # accept 2 values and use the ratio of val1 to val2
+    if val2 is not None:
+        ratio = ratio / val2
+
+    # special case for multiply-by-zero (convert to silence)
+    if ratio == 0:
+        return -float('inf')
+
+    if using_amplitude:
+        return 20 * log(ratio, 10)
+    else:  # using power
+        return 10 * log(ratio, 10)
+
+
+def register_pydub_effect(fn, name=None):
+    """
+    decorator for adding pydub effects to the AudioSegment objects.
+    example use:
+        @register_pydub_effect
+        def normalize(audio_segment):
+            ...
+    or you can specify a name:
+        @register_pydub_effect("normalize")
+        def normalize_audio_segment(audio_segment):
+            ...
+    """
+    if isinstance(fn, basestring):
+        name = fn
+        return lambda fn: register_pydub_effect(fn, name)
+
+    if name is None:
+        name = fn.__name__
+
+    from .audio_segment import AudioSegment
+    setattr(AudioSegment, name, fn)
+    return fn
+
+
+def make_chunks(audio_segment, chunk_length):
+    """
+    Breaks an AudioSegment into chunks that are <chunk_length> milliseconds
+    long.
+    if chunk_length is 50 then you'll get a list of 50 millisecond long audio
+    segments back (except the last one, which can be shorter)
+    """
+    number_of_chunks = ceil(len(audio_segment) / float(chunk_length))
+    return [audio_segment[i * chunk_length:(i + 1) * chunk_length]
+            for i in range(int(number_of_chunks))]
+
+
+def which(program):
+    """
+    Mimics behavior of UNIX which command.
+    """
+    # Add .exe program extension for windows support
+    if os.name == "nt" and not program.endswith(".exe"):
+        program += ".exe"
+
+    envdir_list = [os.curdir] + os.environ["PATH"].split(os.pathsep)
+
+    for envdir in envdir_list:
+        program_path = os.path.join(envdir, program)
+        if os.path.isfile(program_path) and os.access(program_path, os.X_OK):
+            return program_path
+
+
+def get_encoder_name():
+    """
+    Return enconder default application for system, either avconv or ffmpeg
+    """
+    if which("avconv"):
+        return "avconv"
+    elif which("ffmpeg"):
+        return "ffmpeg"
+    else:
+        # should raise exception
+        warn("Couldn't find ffmpeg or avconv - defaulting to ffmpeg, but may not work", RuntimeWarning)
+        return "ffmpeg"
+
+
+def get_player_name():
+    """
+    Return enconder default application for system, either avconv or ffmpeg
+    """
+    if which("avplay"):
+        return "avplay"
+    elif which("ffplay"):
+        return "ffplay"
+    else:
+        # should raise exception
+        warn("Couldn't find ffplay or avplay - defaulting to ffplay, but may not work", RuntimeWarning)
+        return "ffplay"
+
+
+def get_prober_name():
+    """
+    Return probe application, either avconv or ffmpeg
+    """
+    if which("avprobe"):
+        return "avprobe"
+    elif which("ffprobe"):
+        return "ffprobe"
+    else:
+        # should raise exception
+        warn("Couldn't find ffprobe or avprobe - defaulting to ffprobe, but may not work", RuntimeWarning)
+        return "ffprobe"
+
+
+def fsdecode(filename):
+    """Wrapper for os.fsdecode which was introduced in python 3.2 ."""
+
+    if sys.version_info >= (3, 2):
+        PathLikeTypes = (basestring, bytes)
+        if sys.version_info >= (3, 6):
+            PathLikeTypes += (os.PathLike,)
+        if isinstance(filename, PathLikeTypes):
+            return os.fsdecode(filename)
+    else:
+        if isinstance(filename, bytes):
+            return filename.decode(sys.getfilesystemencoding())
+        if isinstance(filename, basestring):
+            return filename
+
+    raise TypeError("type {0} not accepted by fsdecode".format(type(filename)))
+
+
+def get_extra_info(stderr):
+    """
+    avprobe sometimes gives more information on stderr than
+    on the json output. The information has to be extracted
+    from stderr of the format of:
+    '    Stream #0:0: Audio: flac, 88200 Hz, stereo, s32 (24 bit)'
+    or (macOS version):
+    '    Stream #0:0: Audio: vorbis'
+    '      44100 Hz, stereo, fltp, 320 kb/s'
+
+    :type stderr: str
+    :rtype: list of dict
+    """
+    extra_info = {}
+
+    re_stream = r'(?P<space_start> +)Stream #0[:\.](?P<stream_id>([0-9]+))(?P<content_0>.+)\n?(?! *Stream)((?P<space_end> +)(?P<content_1>.+))?'
+    for i in re.finditer(re_stream, stderr):
+        if i.group('space_end') is not None and len(i.group('space_start')) <= len(
+                i.group('space_end')):
+            content_line = ','.join([i.group('content_0'), i.group('content_1')])
+        else:
+            content_line = i.group('content_0')
+        tokens = [x.strip() for x in re.split('[:,]', content_line) if x]
+        extra_info[int(i.group('stream_id'))] = tokens
+    return extra_info
+
+
+def mediainfo_json(filepath, read_ahead_limit=-1):
+    """Return json dictionary with media info(codec, duration, size, bitrate...) from filepath
+    """
+    prober = get_prober_name()
+    command_args = [
+        "-v", "info",
+        "-show_format",
+        "-show_streams",
+    ]
+    try:
+        command_args += [fsdecode(filepath)]
+        stdin_parameter = None
+        stdin_data = None
+    except TypeError:
+        if prober == 'ffprobe':
+            command_args += ["-read_ahead_limit", str(read_ahead_limit),
+                             "cache:pipe:0"]
+        else:
+            command_args += ["-"]
+        stdin_parameter = PIPE
+        file, close_file = _fd_or_path_or_tempfile(filepath, 'rb', tempfile=False)
+        file.seek(0)
+        stdin_data = file.read()
+        if close_file:
+            file.close()
+
+    command = [prober, '-of', 'json'] + command_args
+    res = Popen(command, stdin=stdin_parameter, stdout=PIPE, stderr=PIPE)
+    output, stderr = res.communicate(input=stdin_data)
+    output = output.decode("utf-8", 'ignore')
+    stderr = stderr.decode("utf-8", 'ignore')
+
+    try:
+        info = json.loads(output)
+    except  json.decoder.JSONDecodeError:
+        # If ffprobe didn't give any information, just return it
+        # (for example, because the file doesn't exist)
+        return None
+    if not info:
+        return info
+
+    extra_info = get_extra_info(stderr)
+
+    audio_streams = [x for x in info['streams'] if x['codec_type'] == 'audio']
+    if len(audio_streams) == 0:
+        return info
+
+    # We just operate on the first audio stream in case there are more
+    stream = audio_streams[0]
+
+    def set_property(stream, prop, value):
+        if prop not in stream or stream[prop] == 0:
+            stream[prop] = value
+
+    for token in extra_info[stream['index']]:
+        m = re.match(r'([su]([0-9]{1,2})p?) \(([0-9]{1,2}) bit\)$', token)
+        m2 = re.match(r'([su]([0-9]{1,2})p?)( \(default\))?$', token)
+        if m:
+            set_property(stream, 'sample_fmt', m.group(1))
+            set_property(stream, 'bits_per_sample', int(m.group(2)))
+            set_property(stream, 'bits_per_raw_sample', int(m.group(3)))
+        elif m2:
+            set_property(stream, 'sample_fmt', m2.group(1))
+            set_property(stream, 'bits_per_sample', int(m2.group(2)))
+            set_property(stream, 'bits_per_raw_sample', int(m2.group(2)))
+        elif re.match(r'(flt)p?( \(default\))?$', token):
+            set_property(stream, 'sample_fmt', token)
+            set_property(stream, 'bits_per_sample', 32)
+            set_property(stream, 'bits_per_raw_sample', 32)
+        elif re.match(r'(dbl)p?( \(default\))?$', token):
+            set_property(stream, 'sample_fmt', token)
+            set_property(stream, 'bits_per_sample', 64)
+            set_property(stream, 'bits_per_raw_sample', 64)
+    return info
+
+
+def mediainfo(filepath):
+    """Return dictionary with media info(codec, duration, size, bitrate...) from filepath
+    """
+
+    prober = get_prober_name()
+    command_args = [
+        "-v", "quiet",
+        "-show_format",
+        "-show_streams",
+        filepath
+    ]
+
+    command = [prober, '-of', 'old'] + command_args
+    res = Popen(command, stdout=PIPE)
+    output = res.communicate()[0].decode("utf-8")
+
+    if res.returncode != 0:
+        command = [prober] + command_args
+        output = Popen(command, stdout=PIPE).communicate()[0].decode("utf-8")
+
+    rgx = re.compile(r"(?:(?P<inner_dict>.*?):)?(?P<key>.*?)\=(?P<value>.*?)$")
+    info = {}
+
+    if sys.platform == 'win32':
+        output = output.replace("\r", "")
+
+    for line in output.split("\n"):
+        # print(line)
+        mobj = rgx.match(line)
+
+        if mobj:
+            # print(mobj.groups())
+            inner_dict, key, value = mobj.groups()
+
+            if inner_dict:
+                try:
+                    info[inner_dict]
+                except KeyError:
+                    info[inner_dict] = {}
+                info[inner_dict][key] = value
+            else:
+                info[key] = value
+
+    return info
+
+
+def cache_codecs(function):
+    cache = {}
+
+    @wraps(function)
+    def wrapper():
+        try:
+            return cache[0]
+        except:
+            cache[0] = function()
+            return cache[0]
+
+    return wrapper
+
+
+@cache_codecs
+def get_supported_codecs():
+    encoder = get_encoder_name()
+    command = [encoder, "-codecs"]
+    res = Popen(command, stdout=PIPE, stderr=PIPE)
+    output = res.communicate()[0].decode("utf-8")
+    if res.returncode != 0:
+        return []
+
+    if sys.platform == 'win32':
+        output = output.replace("\r", "")
+
+
+    rgx = re.compile(r"^([D.][E.][AVS.][I.][L.][S.]) (\w*) +(.*)")
+    decoders = set()
+    encoders = set()
+    for line in output.split('\n'):
+        match = rgx.match(line.strip())
+        if not match:
+            continue
+        flags, codec, name = match.groups()
+
+        if flags[0] == 'D':
+            decoders.add(codec)
+
+        if flags[1] == 'E':
+            encoders.add(codec)
+
+    return (decoders, encoders)
+
+
+def get_supported_decoders():
+    return get_supported_codecs()[0]
+
+
+def get_supported_encoders():
+    return get_supported_codecs()[1]
+
+def stereo_to_ms(audio_segment):
+	'''
+	Left-Right -> Mid-Side
+	'''
+	channel = audio_segment.split_to_mono()
+	channel = [channel[0].overlay(channel[1]), channel[0].overlay(channel[1].invert_phase())]
+	return AudioSegment.from_mono_audiosegments(channel[0], channel[1])
+
+def ms_to_stereo(audio_segment):
+	'''
+	Mid-Side -> Left-Right
+	'''
+	channel = audio_segment.split_to_mono()
+	channel = [channel[0].overlay(channel[1]) - 3, channel[0].overlay(channel[1].invert_phase()) - 3]
+	return AudioSegment.from_mono_audiosegments(channel[0], channel[1])
+
@@ -0,0 +1,5 @@
+[wheel]
+universal = 1
+
+[pep8]
+max-line-length = 100
@@ -0,0 +1,42 @@
+__doc__ = """
+Manipulate audio with an simple and easy high level interface.
+
+See the README file for details, usage info, and a list of gotchas.
+"""
+
+from setuptools import setup
+
+setup(
+    name='pydub',
+    version='0.25.1',
+    author='James Robert',
+    author_email='jiaaro@gmail.com',
+    description='Manipulate audio with an simple and easy high level interface',
+    license='MIT',
+    keywords='audio sound high-level',
+    url='http://pydub.com',
+    packages=['pydub'],
+    long_description=__doc__,
+    classifiers=[
+        'Development Status :: 5 - Production/Stable',
+        'License :: OSI Approved :: MIT License',
+        'Programming Language :: Python',
+        'Programming Language :: Python :: 2',
+        'Programming Language :: Python :: 2.7',
+        'Programming Language :: Python :: 3',
+        'Programming Language :: Python :: 3.4',
+        'Programming Language :: Python :: 3.5',
+        'Programming Language :: Python :: 3.6',
+        'Programming Language :: Python :: 3.7',
+        'Programming Language :: Python :: 3.8',
+        'Intended Audience :: Developers',
+        'Operating System :: OS Independent',
+        "Topic :: Multimedia :: Sound/Audio",
+        "Topic :: Multimedia :: Sound/Audio :: Analysis",
+        "Topic :: Multimedia :: Sound/Audio :: Conversion",
+        "Topic :: Multimedia :: Sound/Audio :: Editors",
+        "Topic :: Multimedia :: Sound/Audio :: Mixers",
+        "Topic :: Software Development :: Libraries",
+        'Topic :: Utilities',
+    ]
+)
@@ -0,0 +1,29 @@
+"""
+Recipes are a way to create mixes.
+"""
+
+from abc import ABC, abstractmethod
+from typing import Any, List
+
+class HomepageRoutine(ABC):
+    """
+    A routine creates a row of homepage items.
+    """
+
+    @property
+    @abstractmethod
+    def is_valid(self) -> bool: ...
+
+    def __init__(self) -> None:
+        if not self.is_valid:
+            return
+
+        self.run()
+
+    @abstractmethod
+    def run(self) -> List[Any]:
+        """
+        Creates the homepage items and saves them to the
+        homepage store if self.is_valid is true.
+        """
+        ...
@@ -0,0 +1,38 @@
+from swingmusic.db.userdata import UserTable
+from swingmusic.lib.recipes import HomepageRoutine
+from swingmusic.plugins.mixes import MixesPlugin
+from swingmusic.store.homepage import HomepageStore
+
+
+class ArtistMixes(HomepageRoutine):
+    store_key = "artist_mixes"
+
+    @property
+    def is_valid(self):
+        return MixesPlugin().enabled
+
+    def run(self):
+        users = UserTable.get_all()
+
+        for user in users:
+            mix = MixesPlugin()
+            mixes = mix.create_artist_mixes(user.id)
+
+            if not mixes:
+                continue
+
+            HomepageStore.set_mixes(mixes, entrykey=self.store_key, userid=user.id)
+
+            custom_mixes = []
+            for _mix in mixes:
+                custom_mix = MixesPlugin.get_track_mix(_mix)
+
+                if custom_mix:
+                    custom_mixes.append(custom_mix)
+
+            HomepageStore.set_mixes(
+                custom_mixes, entrykey="custom_mixes", userid=user.id
+            )
+
+    def __init__(self) -> None:
+        super().__init__()
@@ -0,0 +1,40 @@
+from pprint import pprint
+from swingmusic.db.userdata import UserTable
+from swingmusic.lib.recipes import HomepageRoutine
+from swingmusic.lib.recipes.artistmixes import ArtistMixes
+from swingmusic.models.mix import Mix
+from swingmusic.plugins.mixes import MixesPlugin
+from swingmusic.store.homepage import HomepageStore
+
+
+class BecauseYouListened(HomepageRoutine):
+    store_keys = ["because_you_listened_to_artist", "artists_you_might_like"]
+
+    @property
+    def is_valid(self):
+        return MixesPlugin().enabled
+
+    def run(self):
+        users = UserTable.get_all()
+
+        for user in users:
+            entry: dict[str, Mix] = HomepageStore.entries.get(
+                ArtistMixes.store_key
+            ).items.get(user.id)  # type: ignore
+
+            if not entry:
+                continue
+
+            because_you_listened_to_artist, artists_you_might_like = (
+                MixesPlugin().get_because_items(list(entry.values()))
+            )
+
+            if not because_you_listened_to_artist or not artists_you_might_like:
+                continue
+
+            HomepageStore.entries[self.store_keys[0]].items[
+                user.id
+            ] = because_you_listened_to_artist
+            HomepageStore.entries[self.store_keys[1]].items[
+                user.id
+            ] = artists_you_might_like
@@ -0,0 +1,96 @@
+from swingmusic.db.userdata import ScrobbleTable, UserTable
+from swingmusic.lib.home.recentlyadded import get_recently_added_items
+from swingmusic.lib.home.get_recently_played import get_recently_played
+from swingmusic.lib.recipes import HomepageRoutine
+from swingmusic.store.homepage import HomepageStore
+
+
+class RecentlyPlayed(HomepageRoutine):
+    ITEM_LIMIT = 15
+    store_key = "recently_played"
+
+    def __init__(self, userid: int | None = None) -> None:
+        """
+        The userid is provided when we are running this routine
+        outside a cron job. ie. when a user records a new scrobble.
+        """
+        self.userids = [userid] if userid else [user.id for user in UserTable.get_all()]
+
+        # NOTE: When the userid is provided
+        # we need to update the store for that userid only
+        # using the last scrobble entry.
+        self.update_only = userid is not None
+        super().__init__()
+
+    @property
+    def is_valid(self):
+        return True
+
+    def run(self):
+        if self.update_only:
+            last_entry = ScrobbleTable.get_last_entry(self.userids[0])
+
+            if last_entry:
+                items = get_recently_played(
+                    limit=self.ITEM_LIMIT, userid=self.userids[0], _entries=[last_entry]
+                )
+
+                try:
+                    item = items[0]
+                    store_entry = HomepageStore.entries[self.store_key].items[
+                        self.userids[0]
+                    ][0]
+                except IndexError:
+                    store_entry = None
+                    item = None
+
+                if (
+                    store_entry
+                    and item
+                    and store_entry.get("type", "") + store_entry.get("hash", "")
+                    == item.get("type", "") + item.get("hash", "")
+                ):
+                    # If the item is the same as the one in the store
+                    # only update the timestamp
+                    HomepageStore.entries[self.store_key].items[self.userids[0]][0][
+                        "timestamp"
+                    ] = item["timestamp"]
+                else:
+                    # Otherwise, insert the new item
+                    # and remove the oldest item if there are more than 15 items
+                    HomepageStore.entries[self.store_key].items[self.userids[0]].insert(
+                        0, item
+                    )
+
+                    if (
+                        len(
+                            HomepageStore.entries[self.store_key].items[self.userids[0]]
+                        )
+                        > self.ITEM_LIMIT
+                    ):
+                        HomepageStore.entries[self.store_key].items[
+                            self.userids[0]
+                        ].pop()
+
+        for userid in self.userids:
+            items = get_recently_played(limit=self.ITEM_LIMIT, userid=userid)
+            HomepageStore.entries[self.store_key].items[userid] = items
+
+
+class RecentlyAdded(HomepageRoutine):
+    ITEM_LIMIT = 15
+    store_key = "recently_added"
+
+    @property
+    def is_valid(self):
+        return True
+
+    def __init__(self):
+        super().__init__()
+
+    def run(self):
+        items = get_recently_added_items(limit=self.ITEM_LIMIT)
+
+        # NOTE: Recently added is a global entry
+        # So we don't need a userid
+        HomepageStore.entries[self.store_key].items[0] = items
@@ -0,0 +1,83 @@
+from gettext import ngettext
+import pendulum
+
+from swingmusic.crons.cron import CronJob
+from swingmusic.db.userdata import UserTable
+from swingmusic.lib.recipes import HomepageRoutine
+from swingmusic.store.homepage import HomepageStore
+from swingmusic.utils.dates import get_date_range, seconds_to_time_string
+from swingmusic.utils.stats import get_artists_in_period
+
+
+class TopArtists(CronJob, HomepageRoutine):
+    """
+    A routine to populate the top streamed artists/albums in the last week or month
+    """
+
+    hours = 1
+    ITEM_LIMIT = 15
+
+    @property
+    def is_valid(self):
+        """
+        Only valid if it's the middle or last 2 days of this month.
+
+        When the duration is "week", it's valid on the weekend.
+        """
+        if self.duration == "month":
+            now = pendulum.now()
+            middle_day = now.days_in_month // 2
+
+            return (
+                now.day in range(middle_day, middle_day + 2)
+                or now.day > now.days_in_month - 2
+            )
+        if self.duration == "week":
+            return pendulum.now().isoweekday() in (5, 6, 7)
+
+        return False
+
+    def __init__(self, duration: str = "month") -> None:
+        super().__init__()
+        self.duration = duration
+
+        if not self.is_valid:
+            return
+
+    def run(self):
+        if not self.is_valid:
+            self.destroy()
+            return
+
+        self.userids = [user.id for user in UserTable.get_all()]
+
+        for userid in self.userids:
+            date_range = get_date_range(self.duration)
+            artists = get_artists_in_period(date_range[0], date_range[1], userid)[
+                : self.ITEM_LIMIT
+            ]
+
+            artists = [
+                {
+                    "type": "artist",
+                    "hash": artist["artisthash"],
+                    "help_text": seconds_to_time_string(artist["playduration"]),
+                    "secondary_text": str(artist["playcount"])
+                    + " "
+                    + ngettext("play", "plays", artist["playcount"]),
+                }
+                for artist in artists
+            ]
+
+            HomepageStore.entries[f"top_streamed_{self.duration}ly_artists"].items[
+                userid
+            ] = artists
+
+    def destroy(self):
+        """
+        Clear the top streamed entry from the homepage store.
+        """
+        keys = [f"top_streamed_{self.duration}ly_artists"]
+
+        for key in keys:
+            HomepageStore.entries[key].items = {}
@@ -0,0 +1,319 @@
+"""
+This library contains all the functions related to the search functionality.
+"""
+
+from rapidfuzz import process, utils, fuzz
+from unidecode import unidecode
+
+from swingmusic import models
+
+from swingmusic.models.album import Album
+from swingmusic.models.artist import Artist
+from swingmusic.models.playlist import Playlist
+from swingmusic.models.track import Track
+from swingmusic.serializers.album import serialize_for_card as serialize_album
+from swingmusic.serializers.album import serialize_for_card_many as serialize_albums
+from swingmusic.serializers.artist import serialize_for_card, serialize_for_cards
+from swingmusic.serializers.track import serialize_track, serialize_tracks
+
+from swingmusic.store.albums import AlbumStore
+from swingmusic.store.artists import ArtistStore
+from swingmusic.store.tracks import TrackStore
+
+from swingmusic.utils.remove_duplicates import remove_duplicates
+
+# ratio = fuzz.ratio
+# wratio = fuzz.WRatio
+
+
+class Cutoff:
+    """
+    Holds all the default cutoff values.
+    """
+
+    tracks: int = 50
+    albums: int = 50
+    artists: int = 50
+    playlists: int = 50
+
+
+class Limit:
+    """
+    Holds all the default limit values.
+    """
+
+    tracks: int = 150
+    albums: int = 150
+    artists: int = 150
+    playlists: int = 150
+
+
+class SearchTracks:
+    def __init__(self, query: str) -> None:
+        self.query = query
+        self.tracks = TrackStore.get_flat_list()
+
+    def __call__(self, limit: int = Limit.tracks) -> list[models.Track]:
+        """
+        Gets all songs with a given title.
+        """
+
+        track_titles = [unidecode(track.title).lower() for track in self.tracks]
+        results = process.extract(
+            self.query,
+            track_titles,
+            score_cutoff=Cutoff.tracks,
+            limit=limit,
+            processor=utils.default_process,
+            scorer=fuzz.WRatio,
+        )
+
+        tracks: list[Track] = []
+
+        for item in results:
+            track = self.tracks[item[2]]
+            track._score = item[1]
+            tracks.append(track)
+
+        return remove_duplicates(tracks)
+
+
+class SearchArtists:
+    def __init__(self, query: str) -> None:
+        self.query = query
+        self.artists = ArtistStore.get_flat_list()
+
+    def __call__(self, limit: int = Limit.artists):
+        """
+        Gets all artists with a given name.
+        """
+        choices = [unidecode(a.name).lower() for a in self.artists]
+
+        results = process.extract(
+            self.query,
+            choices,
+            score_cutoff=Cutoff.artists,
+            limit=limit,
+            processor=utils.default_process,
+            scorer=fuzz.WRatio,
+        )
+
+        artists: list[Artist] = []
+
+        for item in results:
+            artist = self.artists[item[2]]
+            artist._score = item[1]
+            artists.append(artist)
+
+        return artists
+
+
+class SearchAlbums:
+    def __init__(self, query: str) -> None:
+        self.query = query
+        self.albums = AlbumStore.get_flat_list()
+
+    def __call__(self, limit: int = Limit.albums):
+        """
+        Gets all albums with a given title.
+        """
+
+        choices = [unidecode(a.title).lower() for a in self.albums]
+
+        results = process.extract(
+            self.query,
+            choices,
+            score_cutoff=Cutoff.albums,
+            limit=limit,
+            processor=utils.default_process,
+            scorer=fuzz.token_sort_ratio,
+        )
+
+        albums: list[Album] = []
+
+        for item in results:
+            album = self.albums[item[2]]
+            album._score = item[1]
+            albums.append(album)
+
+        return albums
+
+
+class SearchPlaylists:
+    def __init__(self, playlists: list[models.Playlist], query: str) -> None:
+        self.playlists = playlists
+        self.query = query
+
+    def __call__(self, limit: int = Limit.playlists):
+        choices = [p.name for p in self.playlists]
+        results = process.extract(
+            self.query,
+            choices,
+            score_cutoff=Cutoff.playlists,
+            limit=limit,
+            processor=utils.default_process,
+            scorer=fuzz.WRatio,
+        )
+
+        playlists: list[Playlist] = []
+
+        for item in results:
+            playlist = self.playlists[item[2]]
+            playlist._score = item[1]
+            playlists.append(playlist)
+
+        return playlists
+
+
+_type = models.Track | models.Album | models.Artist
+
+
+def get_titles(items: list[_type]):
+    for item in items:
+        if isinstance(item, models.Track):
+            text = item.og_title
+        elif isinstance(item, models.Album):
+            text = item.title
+        elif isinstance(item, models.Artist):
+            text = item.name
+        else:
+            text = None
+
+        yield text
+
+
+class TopResults:
+    """
+    Joins all tracks, albums and artists
+    then fuzzy searches them as a single unit.
+    """
+
+    @staticmethod
+    def collect_all():
+        all_items: list[_type] = []
+
+        all_items.extend(ArtistStore.get_flat_list())
+        all_items.extend(TrackStore.get_flat_list())
+        all_items.extend(AlbumStore.get_flat_list())
+
+        return all_items, get_titles(all_items)
+
+    @staticmethod
+    def get_track_items(item: Track | Album | Artist, limit=5):
+        tracks: list[Track] = []
+
+        # INFO: If the item is a track, return empty list
+        # to be filled by the results from the top search
+        if isinstance(item, Track):
+            return tracks
+
+        # INFO: If the item is an album, get the tracks from the album
+        if isinstance(item, Album):
+            tracks = TrackStore.get_tracks_by_albumhash(item.albumhash)[:limit]
+            tracks.sort(key=lambda x: x.playduration, reverse=True)
+            return tracks
+
+        # INFO: If the item is an artist, get the tracks from the artist
+        if isinstance(item, Artist):
+            tracks = TrackStore.get_tracks_by_artisthash(item.artisthash)[:limit]
+            tracks.sort(key=lambda x: x.playduration, reverse=True)
+
+        return tracks
+
+    @staticmethod
+    def get_album_items(item: Track | Album | Artist, limit=6):
+        albums: list[Album] = []
+
+        # INFO: If the item is a track or album, search for albums
+        if isinstance(item, Track) or isinstance(item, Album):
+            return albums
+
+        # INFO: If the item is an artist, get the albums from the artist
+        if isinstance(item, Artist):
+            albums = AlbumStore.get_albums_by_artisthash(item.artisthash)[:limit]
+
+        return albums
+
+    @staticmethod
+    def search(
+        query: str,
+        limit: int = None,
+        albums_only=False,
+        tracks_only=False,
+    ):
+        tracks_limit = Limit.tracks if tracks_only else 4
+        albums_limit = Limit.albums if albums_only else limit
+        artists_limit = limit
+
+        # INFO: Individually search all stores as each type has a different scorer
+        tracks = SearchTracks(query)(limit=tracks_limit) if not albums_only else []
+        albums = SearchAlbums(query)(limit=albums_limit)
+        artists = SearchArtists(query)(limit=artists_limit)
+
+        # INFO: Combine all results and sort them by score
+        all_results = artists + tracks + albums
+        all_results = sorted(all_results, key=lambda x: int(x._score), reverse=True)
+
+        # INFO: Get the top result
+        top_result = all_results[0]
+        top_tracks = []
+
+        if not albums_only:
+            top_tracks = TopResults.get_track_items(top_result, limit=tracks_limit)
+
+            # INFO: If there are not enough tracks, fill with search results
+            if len(top_tracks) < tracks_limit:
+                found_tracks_set = {track.trackhash for track in top_tracks}
+
+                for track in tracks:
+                    if track.trackhash not in found_tracks_set:
+                        top_tracks.append(track)
+
+                    if len(top_tracks) >= tracks_limit:
+                        break
+
+            top_tracks = serialize_tracks(top_tracks)
+
+            if tracks_only:
+                return top_tracks
+
+        top_albums = TopResults.get_album_items(top_result, limit=albums_limit)
+
+        # INFO: If there are not enough albums, fill with search results
+        if len(top_albums) < albums_limit:
+            found_albums_set = {album.albumhash for album in top_albums}
+
+            for album in albums:
+                if album.albumhash not in found_albums_set:
+                    top_albums.append(album)
+
+                    if len(top_albums) >= albums_limit:
+                        break
+
+        top_albums = serialize_albums(top_albums)
+
+        if albums_only:
+            return top_albums
+
+        artists = serialize_for_cards(artists)
+
+        if isinstance(top_result, Track):
+            top_result = serialize_track(top_result)
+            top_result["type"] = "track"
+
+        if isinstance(top_result, Album):
+            top_result = serialize_album(top_result)
+            top_result["type"] = "album"
+
+        if isinstance(top_result, Artist):
+            top_result = serialize_for_card(
+                top_result, include={"albumcount", "trackcount"}
+            )
+            top_result["type"] = "artist"
+
+        return {
+            "top_result": top_result,
+            "tracks": top_tracks,
+            "artists": artists,
+            "albums": top_albums,
+        }
@@ -0,0 +1,55 @@
+from itertools import groupby
+import os
+from typing import Callable
+from swingmusic.lib.albumslib import sort_by_track_no
+from swingmusic.models.folder import Folder
+from swingmusic.models.track import Track
+from swingmusic.utils import flatten
+
+
+def sort_tracks(tracks: list[Track], key: str, reverse: bool = False):
+    """
+    Sorts a list of tracks by a key.
+    """
+    if key == "default":
+        return tracks
+
+    sortfunc: Callable[[Track], str] = lambda track: getattr(track, key)
+    if key == "artists" or key == "albumartists":
+        sortfunc = lambda track: getattr(track, key)[0]["name"]
+
+    if key == "disc":
+        # INFO: Group tracks into albums, then sort them by disc number.
+        tracks = sorted(tracks, key=lambda x: x.album.casefold())
+        groups = groupby(tracks, lambda x: x.albumhash)
+
+        return flatten([sort_by_track_no(list(g)) for k, g in groups])
+
+    # INFO: sort tracks by title for a fallback value
+    tracks = sorted(tracks, key=lambda t: t.title.casefold())
+
+    if key == "title" and not reverse:
+        return tracks
+
+    return sorted(
+        tracks,
+        key=lambda track: sortfunc(track).casefold()
+        if isinstance(sortfunc(track), str)
+        else sortfunc(track),
+        reverse=reverse,
+    )
+
+
+def sort_folders(folders: list[Folder], key: str, reverse: bool = False):
+    """
+    Sorts a list of folders by a key.
+    """
+    if key == "default":
+        return folders
+
+    sortfunc: Callable[[Folder], str | float] = lambda folder: getattr(folder, key)
+
+    if key == "lastmod":
+        sortfunc = lambda folder: os.path.getmtime(folder.path)
+
+    return sorted(folders, key=sortfunc, reverse=reverse)
@@ -0,0 +1,332 @@
+import os
+from functools import partial
+from multiprocessing import Pool, cpu_count
+
+from swingmusic import settings
+from swingmusic.config import UserConfig
+from swingmusic.db.libdata import TrackTable
+
+from swingmusic.lib.taglib import extract_thumb, get_tags
+from swingmusic.models.album import Album
+from swingmusic.models.artist import Artist
+from swingmusic.models.track import Track
+from swingmusic.store.folder import FolderStore
+from swingmusic.store.tracks import TrackStore
+from swingmusic.utils import flatten
+from swingmusic.utils.filesystem import run_fast_scandir
+from swingmusic.utils.parsers import get_base_album_title
+from swingmusic.utils.progressbar import tqdm
+from swingmusic.utils.remove_duplicates import remove_duplicates
+
+
+from logging import getLogger
+log = getLogger(__name__)
+
+
+def parse_file_tags(file: str, config: UserConfig) -> dict | None:
+    """Worker function to process individual files"""
+    try:
+        return get_tags(file, config=config)
+    except Exception as e:
+        log.warning(f"Failed to process file {file}: {e}")
+        return None
+
+
+class IndexTracks:
+    def __init__(self) -> None:
+        """
+        Indexes all tracks in the database.
+
+        An instance key is used to prevent multiple instances of the
+        same class from running at the same time.
+        """
+        dirs_to_scan = UserConfig().rootDirs
+
+        if len(dirs_to_scan) == 0:
+            log.warning(
+                (
+                    "The root directory is not configured. "
+                    + "Open the app in your webbrowser to configure."
+                )
+            )
+            return
+
+        try:
+            if dirs_to_scan[0] == "$home":
+                dirs_to_scan = [settings.Paths().USER_HOME_DIR.as_posix()]
+        except IndexError:
+            pass
+
+        files = set()
+
+        for _dir in dirs_to_scan:
+            files = files.union(run_fast_scandir(_dir, full=True)[1])
+
+        unmodified, modified_tracks = self.filter_modded()
+        untagged = files - unmodified
+
+        self.tag_untagged(untagged)
+        self.extract_thumb_with_overwrite(modified_tracks)
+
+    @staticmethod
+    def extract_thumb_with_overwrite(tracks: list[dict[str, str]]):
+        """
+        Extracts the thumbnail from a list of filepaths,
+        overwriting the existing thumbnail if it exists,
+        for modified files.
+        """
+        for track in tracks:
+            try:
+                extract_thumb(
+                    track["filepath"], track["albumhash"] + ".webp", overwrite=True, paths=settings.Paths()
+                )
+            except FileNotFoundError:
+                continue
+
+    @staticmethod
+    def filter_modded():
+        """
+        Removes tracks from the database that have been modified
+        since they were indexed.
+
+        Returns a tuple of unmodified paths and modified tracks.
+        Unmodified paths are indexed and the modified tracks are
+
+        """
+
+        unmodified_paths = set()
+        modified_tracks: list[dict[str, str]] = []
+
+        to_remove = set()
+
+        for track in TrackTable.get_all():
+            try:
+                if track.last_mod == round(os.path.getmtime(track.filepath)):
+                    unmodified_paths.add(track.filepath)
+                    continue
+            except (FileNotFoundError, OSError) as e:
+                log.warning(e)  # REVIEW More informations = good
+                to_remove.add(track.filepath)
+
+            modified_tracks.append(
+                {
+                    "filepath": track.filepath,
+                    "albumhash": track.albumhash,
+                }
+            )
+
+        to_remove = to_remove.union(set(t["filepath"] for t in modified_tracks))
+        TrackTable.remove_tracks_by_filepaths(to_remove)
+
+        # REVIEW: Remove after testing!
+        track = TrackTable.get_tracks_by_filepaths(list(to_remove)[:1])
+        if track:
+            raise Exception("Track not removed")
+        # =============================================================
+
+        return unmodified_paths, modified_tracks
+
+
+    def tag_untagged(self, files: set[str]):
+        config = UserConfig()
+
+        # Create process pool with worker function
+        with Pool(processes=max(1, cpu_count() // 2)) as pool:
+            worker = partial(parse_file_tags, config=config)
+
+            # Process files and track progress
+            results = []
+            for result in tqdm(
+                pool.imap_unordered(worker, files),
+                total=len(files),
+                desc="Reading files",
+            ):
+                if result is not None:
+                    results.append(result)
+
+        # Bulk insert results
+        for tags in results:
+            TrackTable.insert_one(tags)
+            FolderStore.filepaths.add(tags["filepath"])
+
+        print(f"{len(results)} new files indexed")
+        print("Done")
+
+
+
+#
+# Create functions
+#
+
+def create_albums(_trackhashes: list[str] = []) -> list[tuple[Album, set[str]]]:
+    """
+    Creates album objects using the indexed tracks. Takes in an optional
+    list of trackhashes to create the albums from. If no list is provided,
+    all tracks are used.
+
+    The trackhashes are passed when creating albums from the watchdogg module.
+
+    Returns a list of tuples containing the album and the trackhashes in the album.
+    ie:
+
+    >>> list[tuple[Album, set[str]]]
+    """
+    albums = dict()
+
+    if _trackhashes:
+        all_tracks: list[Track] = TrackStore.get_tracks_by_trackhashes(_trackhashes)
+    else:
+        all_tracks: list[Track] = TrackStore.get_flat_list()
+
+    all_tracks = remove_duplicates(all_tracks)
+
+    for track in all_tracks:
+        if track.albumhash not in albums:
+            albums[track.albumhash] = {
+                "albumartists": track.albumartists,
+                "artisthashes": [a["artisthash"] for a in track.albumartists],
+                "albumhash": track.albumhash,
+                "base_title": None,
+                "color": None,
+                "created_date": track.last_mod,
+                "date": track.date,
+                "duration": track.duration,
+                "genres": [*track.genres] if track.genres else [],
+                "og_title": track.og_album,
+                "lastplayed": track.lastplayed,
+                "playcount": track.playcount,
+                "playduration": track.playduration,
+                "title": track.album,
+                "tracks": {track.trackhash},
+                "pathhash": track.pathhash,
+                "extra": {},
+            }
+        else:
+            album = albums[track.albumhash]
+            album["tracks"].add(track.trackhash)
+            album["playcount"] += track.playcount
+            album["playduration"] += track.playduration
+            album["lastplayed"] = max(album["lastplayed"], track.lastplayed)
+            album["duration"] += track.duration
+            album["date"] = min(album["date"], track.date)
+            album["created_date"] = min(album["created_date"], track.last_mod)
+
+            if track.genres:
+                album["genres"].extend(track.genres)
+
+    for album in albums.values():
+        genres = []
+        for genre in album["genres"]:
+            if genre not in genres:
+                genres.append(genre)
+
+        album["genres"] = genres
+        album["genrehashes"] = " ".join([g["genrehash"] for g in genres])
+        album["base_title"], _ = get_base_album_title(album["og_title"])
+
+        del genres
+        trackhashes = album.pop("tracks")
+        album["trackcount"] = len(trackhashes)
+
+        albums[album["albumhash"]] = (Album(**album), trackhashes)
+
+    return list(albums.values())
+
+
+def create_artists( artisthashes: list[str]) -> list[tuple[Artist, set[str], set[str]]]:
+    """
+    Creates artist objects using the indexed tracks. Takes in an optional
+    list of artisthashes to create the artists from. If no list is provided,
+    all tracks are used.
+
+    Returns a list of tuples containing the artist, the trackhashes for the artist
+    and the albumhashes for the artist.
+    ie:
+
+    >>> list[tuple[Artist, set[str], set[str]]]
+    """
+
+    if artisthashes:
+        all_tracks: list[Track] = flatten(
+            [TrackStore.get_tracks_by_artisthash(hash) for hash in artisthashes]
+        )
+    else:
+        all_tracks: list[Track] = TrackStore.get_flat_list()
+
+    all_tracks = remove_duplicates(all_tracks)
+    artists = dict()
+
+    for track in all_tracks:
+        this_artists = [*track.artists]
+
+        for a in track.albumartists:
+            if a not in this_artists:
+                a["in_track"] = False
+                this_artists.append(a)
+
+        for thisartist in this_artists:
+            if thisartist["artisthash"] not in artists:
+                artists[thisartist["artisthash"]] = {
+                    "albumcount": None,
+                    "albums": {track.albumhash},
+                    "artisthash": thisartist["artisthash"],
+                    "created_date": track.last_mod,
+                    "date": track.date,
+                    "duration": track.duration,
+                    "genres": track.genres if track.genres else [],
+                    "name": None,
+                    "names": {thisartist["name"]},
+                    "lastplayed": track.lastplayed,
+                    "playcount": track.playcount,
+                    "playduration": track.playduration,
+                    "trackcount": None,
+                    "tracks": (
+                        {track.trackhash} if thisartist.get("in_track", True) else set()
+                    ),
+                    "extra": {},
+                }
+            else:
+                artist: dict = artists[thisartist["artisthash"]]
+                artist["duration"] += track.duration
+                artist["playcount"] += track.playcount
+                artist["playduration"] += track.playduration
+                artist["albums"].add(track.albumhash)
+                artist["date"] = min(artist["date"], track.date)
+                artist["lastplayed"] = max(artist["lastplayed"], track.lastplayed)
+                artist["created_date"] = min(artist["created_date"], track.last_mod)
+                artist["names"].add(thisartist["name"])
+
+                artist.setdefault("albums", set())
+
+                if thisartist.get("in_track", True):
+                    artist["tracks"].add(track.trackhash)
+
+                if track.genres:
+                    artist["genres"].extend(track.genres)
+
+    for artist in artists.values():
+        artist["albumcount"] = len(artist["albums"])
+        artist["trackcount"] = len(artist["tracks"])
+
+        genres = []
+
+        for genre in artist["genres"]:
+            if genre not in genres:
+                genres.append(genre)
+
+        artist["genres"] = genres
+        artist["genrehashes"] = " ".join([g["genrehash"] for g in genres])
+        artist["name"] = sorted(artist["names"])[0]
+
+        # INFO: Delete temporary keys
+        del artist["names"]
+
+        tracks = artist.pop("tracks")
+        albums = artist.pop("albums")
+
+        # INFO: Delete local variables
+        del genres
+
+        artists[artist["artisthash"]] = (Artist(**artist), tracks, albums)
+
+    return list(artists.values())
@@ -0,0 +1,314 @@
+import pathlib
+from dataclasses import dataclass
+import os
+from io import BytesIO
+from pathlib import Path
+import re
+from typing import Any
+
+import pendulum
+from PIL import Image, UnidentifiedImageError
+from tinytag import TinyTag
+
+from swingmusic.config import UserConfig
+from swingmusic.settings import Defaults, Paths
+from swingmusic.utils.hashing import create_hash
+from swingmusic.utils.parsers import split_artists
+
+
+def parse_album_art(filepath: str):
+    """
+    Returns the album art for a given audio file.
+
+    :params filepath: Path to file
+    :returns: `Pil.Image` if available else None
+    """
+    tags = TinyTag.get(filepath, image=True)
+    image = tags.images.any
+
+    if image:
+        return image.data
+
+    return None
+
+
+def extract_thumb(filepath: str, webp_path: str, overwrite=False, paths:Paths=None) -> bool:
+    """
+    Extracts the thumbnail from an audio file.
+    Returns the path to the thumbnail.
+    """
+    # this function will be run multithreaded.
+    # Modules are not cached in concurrent runs.
+    # If Paths is tried to be imported
+    if paths is None:
+        paths = Paths()
+
+    lg_img_path = paths.lg_thumb_path / webp_path
+    sm_img_path = paths.sm_thumb_path / webp_path
+    xms_img_path = paths.xsm_thumb_path / webp_path
+    md_img_path = paths.md_thumb_path / webp_path
+
+    images = [
+        (lg_img_path, Defaults.LG_THUMB_SIZE),
+        (sm_img_path, Defaults.SM_THUMB_SIZE),
+        (xms_img_path, Defaults.XSM_THUMB_SIZE),
+        (md_img_path, Defaults.MD_THUMB_SIZE),
+    ]
+
+    def save_image(img: Image.Image):
+        width, height = img.size
+        ratio = width / height
+
+        for path, size in images:
+            img.resize((size, int(size / ratio)), Image.LANCZOS).save(path, "webp")
+
+        del img
+
+    if not overwrite and sm_img_path.exists():
+        img_size = os.path.getsize(sm_img_path)
+
+        if img_size > 0:
+            return True
+
+    album_art = parse_album_art(filepath)
+
+    if album_art is not None:
+        try:
+            img = Image.open(BytesIO(album_art))
+        except (UnidentifiedImageError, OSError):
+            return False
+
+        try:
+            save_image(img)
+        except OSError:
+            try:
+                png = img.convert("RGB")
+                save_image(png)
+            except:  # pylint: disable=bare-except
+                return False
+
+        return True
+    return False
+
+
+def parse_date(date_str: str) -> int | None:
+    """
+    Extracts the date from a string and returns a timestamp.
+    """
+    try:
+        date = pendulum.parse(date_str, strict=False)
+        return int(date.timestamp())
+    except Exception as e:
+        return None
+
+
+def clean_filename(filename: str):
+    if "official" in filename.lower():
+        return re.sub(r"\s*\([^)]*official[^)]*\)", "", filename, flags=re.IGNORECASE)
+
+    return filename
+
+
+@dataclass
+class ParseData:
+    artist: str
+    title: str
+    config: UserConfig
+
+    def __post_init__(self):
+        self.artist = split_artists(self.artist, self.config)
+
+
+def extract_artist_title(filename: str, config: UserConfig):
+    """
+    extract data from filename with specified separators
+
+    :params filename: filename
+    :params config: UserConfig for user separators
+    """
+
+    path = Path(filename).with_suffix("")
+
+    path = clean_filename(str(path))
+    split_result = path.split(" - ")
+    split_result = [x.strip() for x in split_result]
+
+    if len(split_result) == 1:
+        return ParseData(
+            "",
+            split_result[0],
+            config,
+        )
+
+    if len(split_result) > 2:
+        try:
+            int(split_result[0])
+
+            return ParseData(
+                split_result[1],
+                " - ".join(split_result[2:]),
+                config,
+            )
+        except ValueError:
+            pass
+
+    artist = split_result[0]
+    title = split_result[1]
+    return ParseData(artist, title, config)
+
+
+def get_tags(filepath: str, config: UserConfig) -> dict:
+    """
+    Parse tags from an audio file.
+    If tag entries are missing, try getting them from the file name
+
+    :param filepath: Path to file.
+    :param config: UserConfig for ``split`` and ``splitignore`` config
+    :return: Metadata dict
+    :raise FileNotFoundError: If filepath is invalid
+    """
+
+    filepath = pathlib.Path(filepath)
+    filename = filepath.stem
+
+    if not filepath.exists():
+        raise FileNotFoundError(filepath)
+
+    last_mod = round(filepath.stat().st_mtime)
+    tags = TinyTag.get(filepath)
+
+    if hasattr(tags, "other"):
+        other = tags.other
+    else:
+        other = {}
+
+    metadata: dict[str, Any] = {
+        "album": tags.album,
+        "albumartists": tags.albumartist,
+        "artists": tags.artist,
+        "title": tags.title,
+        "last_mod": last_mod,
+        "filepath": filepath.as_posix(),
+        "folder": filepath.parent.as_posix(),
+        "bitrate": tags.bitrate,
+        "duration": tags.duration,
+        "track": tags.track,
+        "disc": tags.disc,
+        "genres": tags.genre,
+        "copyright": " ".join(other.get("copyright", [])), # INFO: Extract copyright from extra data
+        "extra": {},
+        "date": parse_date(tags.year or "") or int(last_mod)
+    }
+
+
+    # check the necessary tags and set them
+    no_albumartist: bool = (tags.albumartist == "") or (tags.albumartist is None)
+    no_artist: bool = (tags.artist == "") or (tags.artist is None)
+
+    if no_albumartist and not no_artist:
+        # INFO: If no albumartist, use the artist
+        metadata["albumartists"] = tags.artist
+
+    if no_artist and not no_albumartist:
+        # INFO: If no artist, use the albumartist
+        metadata["artists"] = tags.albumartist
+
+    parse_data = None
+
+    # INFO: If title or album is empty, extract the album and title from the filename
+    to_filename = ["title", "album"]
+    for tag in to_filename:
+        p = metadata[tag]
+        if p == "" or p is None:
+            parse_data = extract_artist_title(filename, config)
+            title = parse_data.title.replace("_", " ")
+            metadata[tag] = title
+
+    # INFO: If artist or albumartist is empty
+    # extract the artist and albumartist from the filename
+    parse = ["artists", "albumartists"]
+    for tag in parse:
+        p = metadata[tag]
+
+        if p == "" or p is None:
+            if not parse_data:
+                parse_data = extract_artist_title(filename, config)
+
+            artist = parse_data.artist
+
+            if artist:
+                metadata[tag] = ", ".join(artist)
+            else:
+                metadata[tag] = "Unknown"
+
+
+    # make values beautiful
+    # INFO: If these are empty, set to "Unknown"
+    to_check = ["album", "albumartists"]
+    for prop in to_check:
+        if not metadata[prop]:
+            metadata[prop] = "Unknown"
+
+    # INFO: Round the bitrate and duration
+    to_round = ["bitrate", "duration"]
+    for prop in to_round:
+        try:
+            metadata[prop] = int(getattr(tags, prop))
+        except TypeError:
+            metadata[prop] = 0
+
+    # INFO: Convert these to int
+    to_int = ["track", "disc"]
+    for prop in to_int:
+        try:
+            metadata[prop] = int(getattr(tags, prop))
+        except (ValueError, TypeError):
+            metadata[prop] = 1
+
+
+    # generate hash
+    # create albumhash using og_album
+    metadata["albumhash"] = create_hash(
+        tags.album or "", metadata.get("albumartists", "")
+    )
+
+    metadata["trackhash"] = create_hash(
+        metadata.get("artists", ""),
+        metadata.get("album", ""),
+        metadata.get("title", ""),
+    )
+
+
+
+    # extract extra information not already in tags
+    extra: dict[str, Any] = {
+        k: v for k, v in tags.as_dict().items() if not k in metadata
+    }
+
+    extra["hashinfo"] = {
+        "algo": "sha1",
+        "format": "[:5]+[-5:]",  # first 5 + last 5 chars
+    }
+
+
+    # REMOVE EMPTY VALUES
+    to_pop = ["filename", "artists", "albumartist", "year"]
+    for key, value in extra.items():
+        # None --bool--> False --not--> True
+        # []   --bool--> False --not--> True
+        # ""   --bool--> False --not--> True
+        # [""] --bool--> True  --not--> False
+
+        if isinstance(value, list) and not "".join(value):
+            to_pop.append(key)
+            continue
+
+        if not value:
+            to_pop.append(key)
+
+    for key in to_pop:
+        extra.pop(key, None)
+
+
+    metadata["extra"] = extra
+    return metadata
@@ -0,0 +1,84 @@
+"""
+This library contains all the functions related to tracks.
+"""
+
+import os
+import pathlib
+
+from swingmusic.lib.pydub.pydub import AudioSegment
+from swingmusic.lib.pydub.pydub.silence import detect_leading_silence, detect_silence
+from swingmusic.utils.threading import ProcessWithReturnValue
+
+
+def get_leading_silence_end(filepath: pathlib.Path):
+    """
+    Returns the leading silence of a track.
+    """
+    format = filepath.suffix.replace(".", "")
+    try:
+        audio = AudioSegment.from_file(filepath, format=format)
+        silence = detect_leading_silence(audio, silence_threshold=-40.0, chunk_size=10)
+    except Exception as e:
+        return 0
+
+    return silence if silence > 1000 else 0
+
+
+def get_trailing_silence_start(filepath: str):
+    """
+    Returns the trailing silence of a track.
+    """
+    format = filepath.suffix.replace(".", "")
+
+    try:
+        audio = AudioSegment.from_file(filepath, format=format)
+        duration = len(audio)
+    except Exception as e:
+        return None
+
+    audio = audio[-30000:] if len(audio) > 30000 else audio
+    silence_groups = detect_silence(audio, silence_thresh=-40.0, seek_step=10)
+
+    if len(silence_groups) == 0:
+        return duration
+
+    silence_group = silence_groups[-1]
+    is_ok = silence_group[1] == len(audio)
+
+    if is_ok:
+        return duration - (silence_group[1] - silence_group[0])
+
+    return duration
+
+
+def get_silence_paddings(ending_file: str, starting_file: str):
+    """
+    Returns the ending silence of a track and the starting silence of the next.
+    """
+    starting_file = pathlib.Path(starting_file)
+    ending_file = pathlib.Path(ending_file)
+
+
+    silence = {"starting_file": 0, "ending_file": 0}
+    ending_thread = None
+    starting_thread = None
+
+    if ending_file.exists():
+        ending_thread = ProcessWithReturnValue(
+            target=get_trailing_silence_start, args=(ending_file,)
+        )
+        ending_thread.start()
+
+    if os.path.exists(starting_file):
+        starting_thread = ProcessWithReturnValue(
+            target=get_leading_silence_end, args=(starting_file,)
+        )
+        starting_thread.start()
+
+    if ending_thread:
+        silence["ending_file"] = ending_thread.join()
+
+    if starting_thread:
+        silence["starting_file"] = starting_thread.join()
+
+    return silence
@@ -0,0 +1,75 @@
+from swingmusic.utils.threading import background
+
+
+import subprocess
+
+
+@background
+def start_transcoding(
+    input_path: str, output_path: str, bitrate: str, container_args: list[str], compression_level: int = 12
+):
+    """
+    Starts a background transcoding process for an audio file.
+
+    This function uses FFmpeg to transcode an audio file from one format to another,
+    with specified bitrate and container format. It runs as a background task.
+
+    Args:
+        input_path (str): The path to the input audio file.
+        output_path (str): The path where the transcoded file will be saved.
+        bitrate (str): The desired bitrate for the output file (e.g., "128k").
+        container_args (list[str]): FFmpeg arguments specific to the output container format.
+        compression_level (int): Compression level (0-9, default: 6).
+
+    Returns:
+        None
+
+    Note:
+        This function is decorated with @background, which means it runs asynchronously.
+        The actual transcoding process is handled by FFmpeg in a subprocess.
+        The function will print status messages about the transcoding process.
+    """
+    # Base command
+    command = [
+        "ffmpeg",
+        "-i",
+        input_path,
+        "-map_metadata", "0",  # Add this line to copy metadata
+        "-b:a",
+        bitrate,
+        "-vn",
+        "-compression_level",
+        str(compression_level),
+        # REVIEW: Idk what any flag below this point does!
+        "-movflags", "faststart+frag_keyframe+empty_moov", # TODO. specify fragment size
+        "-write_xing", "0", # ffmpeg.org/ffmpeg-formats.html
+        "-fflags", "+bitexact", #
+    ]
+
+    # Add format-specific parameters
+    command.extend(container_args)
+
+    # Add output path and overwrite flag
+    command.extend([output_path, "-y"])
+
+    process = subprocess.Popen(
+        command, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL
+    )
+    print(f"Started transcoding process with PID: {process.pid}")
+
+    try:
+        # Wait for the process to complete
+        process.wait()
+        print(f"Transcoding process (PID: {process.pid}) completed")
+    except KeyboardInterrupt:
+        print(f"Transcoding interrupted. Terminating process (PID: {process.pid})")
+    finally:
+        # Ensure the process is terminated
+        try:
+            process.terminate()
+            process.wait(timeout=5)  # Wait up to 5 seconds for graceful termination
+        except subprocess.TimeoutExpired:
+            print(
+                f"Process (PID: {process.pid}) did not terminate gracefully. Killing..."
+            )
+            process.kill()
@@ -0,0 +1,372 @@
+"""
+This library contains the classes and functions related to the watchdog file watcher.
+"""
+
+import json
+import os
+import sqlite3
+import time
+
+from watchdog.events import PatternMatchingEventHandler
+from watchdog.observers.api import BaseObserverSubclassCallable
+from watchdog.observers import Observer
+
+from swingmusic import settings
+from swingmusic.config import UserConfig
+from swingmusic.db.libdata import TrackTable
+from swingmusic.db.userdata import LibDataTable
+from swingmusic.lib.colorlib import process_color
+from swingmusic.lib.tagger import create_albums, create_artists
+from swingmusic.lib.taglib import extract_thumb, get_tags
+from swingmusic.logger import log
+from swingmusic.models import Artist, Track
+from swingmusic.store.albums import AlbumStore
+from swingmusic.store.artists import ArtistMapEntry, ArtistStore
+from swingmusic.store.tracks import TrackStore
+
+
+class Watcher:
+    """
+    Contains the methods for initializing and starting watchdog.
+    """
+
+    observers: list[BaseObserverSubclassCallable] = []
+
+    def __init__(self):
+        self.observer = Observer()
+
+    def run(self):
+        """
+        Starts watchers for each dir in root_dirs
+        """
+
+        trials = 0
+
+        while trials < 10:
+            try:
+                # dirs = sdb.get_root_dirs()
+                dirs = UserConfig().rootDirs
+                dirs = [rf"{d}" for d in dirs]
+
+                dir_map = [
+                    {"original": d, "realpath": os.path.realpath(d)} for d in dirs
+                ]
+                break
+            except sqlite3.OperationalError:
+                trials += 1
+                time.sleep(1)
+        else:
+            log.error(
+                "WatchDogError: Failed to start Watchdog. Waiting for database timed out!"
+            )
+            return
+
+        if len(dirs) == 0:
+            log.warning(
+                "WatchDogInfo: No root directories configured. Watchdog not started."
+            )
+            return
+
+        dir_map = [d for d in dir_map if d["realpath"] != d["original"]]
+
+        # if len(dirs) > 0 and dirs[0] == "$home":
+        #     dirs = [settings.USER_HOME_DIR]
+
+        if any([d == "$home" for d in dirs]):
+            dirs = [settings.Paths().USER_HOME_DIR]
+
+        event_handler = Handler(root_dirs=dirs, dir_map=dir_map)
+
+        for _dir in dirs:
+            exists = os.path.exists(_dir)
+
+            if not exists:
+                log.error("WatchdogError: Directory not found: %s", _dir)
+
+        for _dir in dirs:
+            self.observer.schedule(
+                event_handler, os.path.realpath(_dir), recursive=True
+            )
+            self.observers.append(self.observer)
+
+        try:
+            self.observer.start()
+            log.info("Started watchdog")
+        except (FileNotFoundError, PermissionError):
+            log.error(
+                "WatchdogError: Failed to start watchdog,  root directories could not be resolved."
+            )
+            return
+        except OSError as e:
+            log.error("Failed to start watchdog. %s", e)
+            return
+
+        try:
+            while True:
+                time.sleep(1)
+        except KeyboardInterrupt:
+            self.stop_all()
+
+        for obsv in self.observers:
+            obsv.join()
+
+    def stop_all(self):
+        """
+        Unschedules and stops all existing watchers.
+        """
+        log.info("Stopping all watchdog observers")
+        for obsv in self.observers:
+            obsv.unschedule_all()
+            obsv.stop()
+
+    def restart(self):
+        """
+        Stops all existing watchers, refetches root_dirs from the db
+        and restarts the watchers.
+        """
+        log.info("🔃 Restarting watchdog")
+        self.stop_all()
+        self.run()
+
+
+def handle_color(albumhash: str):
+    entry = LibDataTable.find_one(albumhash, "album")
+
+    if entry and entry.color:
+        return
+
+    colors = process_color(albumhash, is_album=True)
+
+    if colors:
+        return
+
+    if entry is None:
+        LibDataTable.insert_one(
+            {"itemhash": albumhash, "color": colors[0], "itemtype": "album"}
+        )
+    else:
+        LibDataTable.update_one(albumhash, {"color": colors[0]})
+
+    return colors
+
+
+def add_track(filepath: str) -> None:
+    """
+    Processes the audio tags for a given file ands add them to the database and store.
+
+    Then creates the folder, album and artist objects for the added track and adds them to the store.
+    """
+
+    TrackStore.remove_track_by_filepath(filepath)
+
+    config = UserConfig()
+    tags = get_tags(filepath, config)
+
+    # if the track is somehow invalid, return
+    if tags is None or tags["bitrate"] == 0 or tags["duration"] == 0:
+        return
+
+    TrackTable.insert_one(tags)
+    extract_thumb(filepath, tags["albumhash"] + ".webp", overwrite=True)
+
+    colors = handle_color(tags["albumhash"])
+    track = Track(**tags)
+    TrackStore.add_track(track)
+
+    # SECTION: Index album
+    albumentry = AlbumStore.albummap.get(track.albumhash)
+
+    if albumentry is None:
+        album, trackhashes = create_albums([track.trackhash])[0]
+        AlbumStore.index_new_album(album, trackhashes)
+    else:
+        trackhash_exists = track.trackhash in albumentry.trackhashes
+
+        if not trackhash_exists:
+            albumentry.trackhashes.add(track.trackhash)
+            albumentry.album.trackcount += 1
+            albumentry.set_color(colors[0]) if colors else None
+
+    # SECTION: Index artist
+    artists = create_artists(track.artisthashes)
+
+    for artist in artists:
+        ArtistStore.artistmap[artist[0].artisthash] = ArtistMapEntry(
+            artist=artist[0],
+            albumhashes=artist[1],
+            trackhashes=artist[2],
+        )
+
+
+def remove_track(filepath: str) -> None:
+    """
+    Removes a track from the music dict.
+    """
+    try:
+        track = TrackStore.get_tracks_by_filepaths([filepath])[0]
+    except IndexError:
+        return
+
+    db.remove_tracks_by_filepaths(filepath)
+    TrackStore.remove_track_by_filepath(filepath)
+
+    empty_album = TrackStore.count_tracks_by_trackhash(track.albumhash) > 0
+
+    if empty_album:
+        AlbumStore.remove_album_by_hash(track.albumhash)
+
+    artists: list[Artist] = track.artists + track.albumartists  # type: ignore
+
+    for artist in artists:
+        empty_artist = not ArtistStore.artist_has_tracks(artist.artisthash)
+
+        if empty_artist:
+            ArtistStore.remove_artist_by_hash(artist.artisthash)
+
+
+class Handler(PatternMatchingEventHandler):
+    files_to_process = []
+    files_to_process_windows = []
+    file_sizes = {}
+
+    root_dirs = []
+    dir_map = []
+
+    def __init__(self, root_dirs: list[str], dir_map: dict[str:str]):
+        self.root_dirs = root_dirs
+        self.dir_map = dir_map
+        patterns = [f"*{f}" for f in settings.SUPPORTED_FILES]
+
+        PatternMatchingEventHandler.__init__(
+            self,
+            patterns=patterns,
+            ignore_directories=True,
+        )
+
+    def get_abs_path(self, path: str):
+        """
+        Convert a realpath to a path relative to the matching root directory.
+        """
+        for d in self.dir_map:
+            if d["realpath"] in path:
+                return path.replace(d["realpath"], d["original"])
+
+        return path
+
+    def on_created(self, event):
+        """
+        Fired when a supported file is created.
+        """
+        try:
+            self.file_sizes[event.src_path] = os.path.getsize(event.src_path)
+        except FileNotFoundError:
+            return
+
+        self.files_to_process.append(event.src_path)
+        self.files_to_process_windows.append(event.src_path)
+
+    def on_deleted(self, event):
+        """
+        Fired when a delete event occurs on a supported file.
+        """
+        path = self.get_abs_path(event.src_path)
+        remove_track(path)
+
+    def on_moved(self, event):
+        """
+        Fired when a move event occurs on a supported file.
+        """
+        trash = "share/Trash"
+
+        if trash in event.dest_path:
+            path = self.get_abs_path(event.src_path)
+            remove_track(path)
+
+        elif trash in event.src_path:
+            path = self.get_abs_path(event.dest_path)
+            add_track(path)
+
+        elif trash not in event.dest_path and trash not in event.src_path:
+            dest_path = self.get_abs_path(event.dest_path)
+            src_path = self.get_abs_path(event.src_path)
+
+            add_track(dest_path)
+            remove_track(src_path)
+
+    def on_closed(self, event):
+        """
+        Fired when a created file is closed.
+        NOT FIRED IN WINDOWS
+        """
+        try:
+            # Get initial file size
+            initial_size = os.path.getsize(event.src_path)
+
+            # Wait for 10 seconds
+            time.sleep(10)
+
+            # Check if file size has changed
+            current_size = os.path.getsize(event.src_path)
+
+            if current_size > 0 and current_size == initial_size:
+                path = self.get_abs_path(event.src_path)
+                add_track(path)
+                # Remove from processing list only after successful processing
+                self.files_to_process.remove(event.src_path)
+            else:
+                # File is still being modified or has been deleted
+                log.info(
+                    f"File {event.src_path} is still being modified. Skipping processing for now."
+                )
+        except FileNotFoundError:
+            # file was closed and deleted.
+            log.info(f"File {event.src_path} was closed and deleted before processing.")
+        except ValueError:
+            # file was already removed from the list by another event handler.
+            log.info(
+                f"File {event.src_path} was already removed from the processing list."
+            )
+
+    def on_modified(self, event):
+        # this event handler is triggered twice on windows
+        # for copy events. We need to test how this behaves in
+        # Linux.
+
+        if event.src_path not in self.files_to_process_windows:
+            return
+
+        # Check if file write operation is complete
+        try:
+            current_size = os.path.getsize(event.src_path)
+        except FileNotFoundError:
+            # File was deleted or moved
+            return
+
+        previous_size = self.file_sizes.get(event.src_path, -1)
+
+        if current_size == previous_size:
+            # Wait for a short duration to ensure the file write operation is complete
+            time.sleep(10)
+
+            # Check the file size again
+            try:
+                current_size = os.path.getsize(event.src_path)
+            except FileNotFoundError:
+                # File was deleted or moved
+                return
+
+            if current_size == previous_size:
+                try:
+                    os.rename(event.src_path, event.src_path)
+                    path = self.get_abs_path(event.src_path)
+                    remove_track(path)
+                    add_track(path)
+                    self.files_to_process_windows.remove(event.src_path)
+                    del self.file_sizes[event.src_path]
+                except OSError:
+                    # File is locked
+                    pass
+                return
+
+        # Update the file size for the next iteration
+        self.file_sizes[event.src_path] = current_size