LMS - Lightweight Music Server
LMS is a self-hosted music streaming software: access your music collection from anywhere using a web interface!
A demo instance is available. Note the administration settings are not available.
Main features
- Subsonic/OpenSubsonic API support
- Multi-valued tags:
genre, mood, artists, …
- Artist relationships:
composer, conductor, lyricist, mixer, performer, producer, remixer
- Release types:
album, single, EP, compilation, live, …
- Release groups support to show different versions of albums, such as remasters, reissues, etc.
- MusicBrainz Identifier support to handle duplicated artist and release names
- ListenBrainz support for:
- Scrobbling and synchronizing listens
- Synchronizing ‘love’ feedbacks
- Last.fm support for scrobbling listens
- Recommendation engine
- Multi-library support
- ReplayGain support
- Audio transcoding for compatibility and reduced bandwidth
- User management, with several authentication backends
- Podcasts support
- Playlists support
- Lyrics support
- Jukebox support
Music discovery
LMS provides several ways to help you find the music you like:
- Tag-based filters (ex: “Rock”, “Metal and Aggressive”, “Electronic and Relaxed”, …)
- Recommendations for similar artists and albums
- Radio mode, with endless filling of the play queue with tracks similar to what is there
- Searches in album, artist and track names (including sort names)
- Starred Albums/Artists/Tracks
- Random/Starred/Most played/Recently played/Recently added for Artist/Albums/Tracks, allowing you to search for things like:
- Recently added Electronic artists
- Random Metal and Aggressive albums
- Most played Relaxed tracks
- Starred Jazz albums
- …
Note: the recommendation engine supports two modes:
- Audio similarity (default): uses MusicNN embeddings. Extraction throughput depends heavily on both the number of scanner threads and the speed of your storage. Typical reported speeds range from ~1k tracks/hour on a Raspberry Pi 4 to ~25k tracks/hour on an Intel Core i5-13500.
- Tag-based: uses metadata tags
genre, mood, grouping, and language.
LMS primarily relies on tags to organize your music collection but also supports browsing by directory using the Subsonic/OpenSubsonic API.
LMS supports an Artist information folder to manage metadata and images for artists. This folder can be placed anywhere within the scanned libraries but is best located in a dedicated ArtistInfo directory in the root of a media library for maximum compatibility with other softwares.
The folder must follow a structure defined by Kodi, as detailed here. artist.nfo files are used to define additional artist information such as biography, sort name, and MusicBrainz ArtistID. See the format here.
The canonical artist name used by LMS is the one specified in the artist.nfo file; If an artist.nfo exists but does not provide a name, the name of the containing folder is used. The same goes for the sort name, when the artist.nfo file provides one.
Otherwise, LMS picks the artist name and sort name found in the most recent album artist tags, then in the most recent artist tags of any kind, tags matched using a MusicBrainz ArtistID first. Artists without a MusicBrainz ArtistID keep their name, as it is what identifies them.
Excluding files from scan
Place a .lmsignore file at the root of a media library to exclude files or directories from scanning.
The file uses a gitignore-inspired reduced syntax:
*.jpg: ignore all .jpg files at any depth/Unsorted/: ignore the top-level Unsorted/ directory onlyextras/: ignore any extras/ directory at any depth!cover.jpg: re-include a file previously matched by a broader rule?: matches any single character except /[abc]: character class
Lines starting with # are comments. An empty file has no effect.
Note: only one .lmsignore file per library root is supported; files placed in subdirectories are ignored.
Filtering
It is possible to apply global filters on your collection using genre, mood, grouping, language, codec, and by music library. More tags, including custom ones, can be added in the database administration settings.
Note: You can use the lms-audioinfo tool to get an idea of the tags parsed by LMS.
Multiple artists
LMS works best when using the default Picard settings, where the artist tag contains a single display-friendly value, and the artists tag holds the actual artist names. This ensures a cleaner, more organized representation of artist names, when multiple artists are involved.
Multiple album artists
While LMS can manage multiple album artists using the albumartist tag, it works better when using the custom albumartists and albumartistssort tags, similar to how it handles regular artist tags.
Note: if you use Picard, add this script to set up both artist and album artist tags:
$setmulti(artistssort,%_artists_sort%)
$setmulti(albumartists,%_albumartists%)
$setmulti(albumartistssort,%_albumartists_sort%)
Extended artist and MusicBrainz ID support
LMS supports several non-standard tags to allow more accurate artist identification:
- MusicBrainz identifiers for artist relationships:
musicbrainz_composerid, musicbrainz_conductorid, musicbrainz_lyricistid, musicbrainz_mixerid, musicbrainz_producerid, musicbrainz_remixerid
- Sort order variants: ,
albumartistssort, composerssort, conductorssort, lyricistssort, mixerssort, producerssort, remixerssort. Singular forms of these tags are also accepted (e.g., conductorsort, lyricistsort, etc.)
Album track grouping
The recommended way to group tracks within an album is to use the musicbrainz_albumid tag.
When this tag is not present, LMS will attempt to group them as best as possible: if the analyzed file contains a disc number and the total number of discs is greater than 1, sibling directories are also scanned to find a matching album.
Otherwise, LMS will only consider albums within the current directory.
For an album to be considered a match, the following conditions must be met:
- Same name
- Same total number of discs
- Identical ‘compilation’ flag value
- Same record labels
- Same barcode
Artwork
LMS scans and serves artwork in jpg/jpeg, png, bmp, webp and gif formats, both as standalone files and embedded in track tags. webp and gif images are always served at their original size: they are never resized, even when a client requests a specific size.
Album image lookup
The recommended method is to name the album image file using the release’s MusicBrainz ReleaseID. This file can be placed anywhere within one of the scanned libraries.
If no file with the MusicBrainz ReleaseID is found, LMS will search for a file named cover, front, folder or default (or other names configured in lms.conf, see the cover-preferred-file-names setting), using this logic:
- Identify the album’s directory: for multi-disc albums, LMS determines the longest common path among all the disc directories (e.g. a
cover.jpg placed alongside the CD1/CD2 folders).
- Fallback search: if no image is found there, LMS will then search within each individual disc folder.
If still no image is found, LMS falls back to images embedded in the album’s tracks, preferring an embedded “Front Cover” image, then “Media”, then “Other” (as some tracks may be badly tagged), using the earliest disc/track that has one, and the largest image as a tiebreaker.
Disc image lookup
LMS automatically associates images with each disc in your collection. Name the image file after the disc’s subtitle or another identifier configured in lms.conf (see the medium-image-file-names setting), and place it in the same directory as the disc’s tracks. If no suitable image is found, LMS will also look for embedded images within the tracks of the disc.
Artist image lookup
The recommended method is to name the artist image file using the artist’s MusicBrainz ArtistID. This file can be placed anywhere within one of the scanned libraries.
If no file with the MusicBrainz ArtistID is found, LMS will look for a file named thumb, folder or fanart (or other names configured in lms.conf, see the artist-info-image-file-names setting) in the artist information directory, where the corresponding artist.nfo file is located.
If neither exists, it will then search for a file named artist (or another name configured in lms.conf, see the artist-image-file-names setting) in the artist’s directories, using this logic:
- Identify the artist’s directory: LMS selects all albums by the artist using the “AlbumArtist” link and determines the longest common path among them.
- Scan for the image: the directory is scanned starting from this common path, moving upwards if needed, until the artist image file is found.
- Fallback search: if no image is found, LMS will then search within each individual album folder.
If still no image is found, LMS falls back to the preferred artwork of the artist’s earliest release (by original release date). This behavior is enabled by default and can be disabled in the admin Scan Settings.
Playlist support
LMS supports playlist files in m3u and m3u8 formats. These playlists are synced during the scan process and are available as public shared playlists.
Playlist image lookup
LMS resolves a playlist’s cover image in this order:
#EXTIMG: directive. An explicit path declared inside the playlist file (URLs are ignored).- Same-name image file. An image file in the same directory as the playlist that shares the playlist’s filename stem.
- First track’s artwork – if neither of the above is found, the artwork of the first track in the playlist is used as a fallback.
Lyrics support
LMS supports lyrics in lrc files, txt files, and embedded track metadata. Both synchronized and unsynchronized lyrics are supported.
Keyboard shortcuts
- Play/pause: Space
- Previous track: Ctrl + Left
- Next track: Ctrl + Right
- Decrease volume: Ctrl + Down
- Increase volume: Ctrl + Up
- Seek back by 5 seconds: Ctrl + Shift + Left
- Seek forward by 5 seconds: Ctrl + Shift + Right
Installation
See INSTALL.md file.
Contributing
Any feedback is welcome:
- feel free to participate in discussions if you have questions,
- report any bug or request for new features in the issue tracker,
- submit your pull requests based on the develop branch.
评论