AudioDistillery

Still — a non-destructive workshop for turning long audio recordings (concerts, vinyl rips, DJ mixes, multitrack field recordings) into finished, mastered, tagged albums — exported to WAV / FLAC / MP3 / AAC.

More screenshots

Why this exists

I play in a band, and every rehearsal ends the same way: a long continuous recording — sometimes a single stereo file, sometimes the separate inputs of a portable multitrack recorder — that nobody ever splits into songs, because doing it with the existing tools is tedious. Audacity means manual labels and destructive edits; WaveLab does everything, but at the price of a steep learning curve and, in our experience, real tagging bugs (see our reports on a malformed trkn atom that makes track numbers invisible in Apple Music and a lost disc number written as 0/0).

AudioDistillery is the tool I wanted instead: load the session (multitrack or not), let it detect what looks like songs, turn them into a track list in a couple of clicks, name them, rebalance the inputs, optionally tag the lot ("rehearsal of March 12th, with the band…" — cover art included), and export to the usual formats with each format's native tags written correctly. Nothing outrageously complicated, nothing destructive, and the tagging bugs stop at our door: the exported files are checked against the pickiest consumer (Apple Music) rather than against a spec reading.

To be clear, none of this is meant as condescending criticism: those applications do a staggering number of very hard things, and I understand both how difficult they are to maintain and that their workflow simply isn't the one that fits me. AudioDistillery only tries to nail one narrow job — and I hope it can help other amateur musicians at the end of a rehearsal too. :)

• Load WAV, FLAC, MP3, M4A or AIFF (multi-GB files stream — nothing is loaded whole into RAM).
• Navigate a fast multi-resolution waveform (wheel = zoom at cursor, minimap).
• Mark each track as a start/end region: press M at the start then at the end, or drag a selection and turn it into a track. Everything outside the regions (applause, gaps, lead-in) is simply ignored at export. Silence detection can propose the regions for you.
• Mix time-synchronized multitrack layers (e.g. the separate inputs of a field recorder): per-layer gain/mute/solo, per-track overrides, chained takes.
• Master in real time with your own plugins — Audio Units and VST3, with native plugin editors, at three scopes: per layer, per track (active only inside that track), and on the master bus. Save chains as presets; exports render through the exact same signal path, latency-compensated.
• Tag the album: metadata, multi-disc numbering, filename macros, cover art.
• Rename tracks inline, then export sample-accurate cuts in parallel with per-track progress and a final report.
• Save the whole session as a tiny .still project file.
• Pick your look: Alambic (default), Alambic Light, Signal or Atelier.

Your source files are never touched. They are opened read-only; exporting only ever creates new files. Delete the app and the project file and your sources are byte-for-byte intact (this is enforced by an integration test).

Requirements

None: a static FFmpeg is bundled with the app (fetched at build time by scripts/fetch-ffmpeg.mjs). Resolution order at runtime: the STILL_FFMPEG env var if set, then the bundled sidecar, then a system-installed ffmpeg (Homebrew/PATH) as fallback.

Development

npm install
npm run tauri dev          # run the app
npm test                   # frontend unit tests
cd src-tauri && cargo test # backend tests (core logic + integration)
npm run tauri build        # produce installers

Architecture, command contract and design system:

• ARCHITECTURE.md — the architecture reference (signal flow, the two rules, plugin hosting).
• CLAUDE.md — the two non-negotiable rules and their decisive tests.
• src-tauri/COMMANDS.md — frontend/backend contract.
• design/DESIGN.md — design system ("Alambic" direction).

Contributing

See CONTRIBUTING.md — start with the two non-negotiable rules; they are what makes this codebase safe to change. Notable releases are listed in CHANGELOG.md.

License

AudioDistillery is released under the MIT License.

Third-party notices

• VST is a trademark of Steinberg Media Technologies GmbH, registered in Europe and other countries. VST 3 hosting is implemented on the MIT-licensed vst3 Rust bindings (coupler-rs); the VST 3 SDK itself is MIT-licensed by Steinberg since version 3.8.
• Audio Units and macOS are technologies of Apple Inc.
• Windows builds may include ASIO Interface Technology by Steinberg Media Technologies GmbH (ASIO is a trademark of Steinberg Media Technologies GmbH). The ASIO SDK is fetched from Steinberg at build time and is never redistributed by this repository.
• Exports are performed by a bundled FFmpeg binary (separate sidecar process), built and distributed by ffmpeg.martin-riedl.de under the GPL; its sources are available from that site. FFmpeg is a trademark of Fabrice Bellard.
• Audio decoding uses Symphonia (MPL-2.0, file-level copyleft — consumed unmodified as a registry dependency). All other Rust and npm dependencies are under permissive licenses (MIT / Apache-2.0 / ISC); see each package's metadata.