CATALOGUE · ENRICH · ARCHIVE
A library for the
music you actually
collect.
MusicHoarder ingests your messy folders, identifies every track by acoustic fingerprint, pulls metadata from nine open sources, and writes a tidy, deduplicated library to disk — artist / album (year) / — the way it should be.
Open source · MIT · runs on your own hardware
live pipeline ~/Downloads/music_dump_2024
[scan] discovered 47 new audio files
[fp] AcoustID match (0.94) → 3f9e8c72-1a4b…
[meta] MusicBrainz: Radiohead — In Rainbows — "Nude"
[art] CAA: fetched front-1500.jpg (214 KB)
[lyr] LRCLIB: synced lyrics (142 lines)
[dupe] duplicate → keeping FLAC over 320 MP3
[write] → /dest/Radiohead/In Rainbows (2007)/03 Nude.flac
[fp] low confidence (0.62) — flagged for review
processed 8,955 remaining 3,892 eta 00:14:32
01
acoustic fingerprinting
Chromaprint + AcoustID identifies tracks even when filename is "track_047.mp3" and the embedded tags are blank.
02
nine enrichment sources
MusicBrainz, Discogs, Spotify, Last.fm, AcoustID, Apple Music, LRCLIB, Genius, Cover Art Archive — fallback chain, never just one.
03
deduplication
Bit-aware dedupe keeps the highest-quality copy; logs every decision so you can audit later.
04
manual review queue
Anything under your confidence threshold lands in a queue. You decide.
05
original files, untouched
MusicHoarder reads but never destroys. The source folder is yours to delete or keep.
06
fully open formats
Library writes are plain files in plain folders. No proprietary database, no lock-in.
THE PIPELINE
Seven stages. Streamed end-to-end.
Scan
walk source tree, sniff format
Fingerprint
Chromaprint hash → AcoustID
Lookup
MusicBrainz + Discogs + Spotify
Artwork
CAA → Last.fm → Spotify upscale
Lyrics
LRCLIB synced → Genius fallback
Dedupe
compare bitrate + length + hash
Write
rename, tag, sort, move
SOURCES
Imports what you actually have.
Local folders
~/Music · ~/Downloads · external drives · NAS shares (SMB/AFP)
Spotify
Liked songs + playlists, queued for purchase or fingerprint-matched against your library
MusicBrainz / Discogs
Open metadata catalogues — the gold standard.
AcoustID / Chromaprint
Acoustic fingerprints. Identifies the recording, not the filename.