We believe that your creative work should never die in darkness — whether the darkness is between you and your past self, or between your work and the person who needs it.
MusicCat catalogues the music you've already written — full DAW projects, voice memos, chord sketches, finished mixes — and listens to the audio to surface what's in it: tempo, key, mood, transcribed lyrics, even the music theory under your fingers. AI as a mirror, not a generator: it describes what you've done so a song you wrote at 2 AM three years ago can find its way back to you. Built for anyone who writes music — in a DAW, on a guitar, or with a melody on the bus.
The client watches the folders you pick and notices every project file — Ableton, Cubase, Bitwig, MPC, Reaper, dawproject — plus standalone audio: voice-memo melodies and finished mixes.
An LLM reads the stems, parses the DAW file, and writes a searchable description — genre, mood, BPM, key, instrumentation.
Filter, sort, and preview. Jump back to the session in your DAW when you find the track you half-remember.
Invite a trusted collaborator by email. They get their own "Shared with me" view and can leave timestamped comments.
Your projects and audio remain yours. We don’t sell your data or share your catalogue with third parties.
We never use your uploads to train or fine-tune any model.
Anonymous usage events only — no third-party trackers, no session recording, no audio leaving your machine for anything other than the catalogue you asked us to build.
Every project at a glance — BPM, key, duration, time signature, AQ score, mood and genre tags, mixdown preview, lock state, and a one-click jump back into the DAW.
Filter by DAW, genre, mood, key, BPM range, heaviness, and project status. Sort by what you actually look for — last worked on, most recently described, AQ score.
The catalogue surfaces what the DAW file knows (tempo, key, time signature, plugins) alongside librosa-derived audio features (energy, brightness, danceability, harmonic vs. percussive, loudness, rolloff) — both feed the AI description.
Not every idea starts in a DAW. Hum a melody into your phone, jot down a chord loop, capture a chorus on the bus — Song Seeds are first-class catalogue entries with full audio analysis (BPM, key, energy) and Whisper-transcribed lyrics. The song doesn’t die between voice memo and DAW.
Spoken thoughts about other tracks, mixing reference recordings, lyric ideas you sing to yourself — Audio Notes capture audio that isn’t a song idea on its own. Whisper transcribes them so you can search the words you said weeks ago.
If there are vocals in the mix, MusicCat transcribes them automatically with Whisper and stores the lyrics alongside the track. Edit the transcript when you want the real released lyrics instead. Lyrics become searchable text — "the song with ‘midnight train’ in the second verse" becomes a one-second lookup.
A music-theory IDE built into the catalogue. Plug a MIDI keyboard in, play any chord, and Explorer names it, shows every inversion, every scale it lives in, every reasonable next chord, and every modulation it opens up. Passive reference, not generative AI — it explains what you’re hearing, you stay the composer.
The desktop client is what scans your DAW folders and syncs to your server. Current release: v0.3.18.