midi-transcribe
Skill16 starsChanged 37 days agoMITView on GitHub
Transcribe an audio stem (e.g. an isolated sax or synth line) to a MIDI file using Spotify basic-pitch with tuned parameters. Use when you need a MIDI version of a monophonic or lightly polyphonic audio part to edit in Ableton. Experimental; not benchmarked against Ableton's built-in audio-to-MIDI.
In this repository8 other files it gives its agents.
- engine/CLAUDE.mdCLAUDE.md
- als-buildSkill
- als-filesSkill
- als-warpSkill
- engineSkill
- midi-compareSkill
- stem-verifySkill
- tempo-driftSkill
The file
Files
You transcribe audio to MIDI using the bundled Ableton engine (the ableton
command on PATH; see the engine skill for the full command reference and
fallback paths). The heavy basic-pitch dependency is installed on demand by
the dispatcher — no setup needed.
Experimental. This uses Spotify basic-pitch. It has not been benchmarked against Ableton's built-in audio-to-MIDI conversion and is not verified to do better; treat the output as a rough starting point to correct by ear. Tell the user this when you produce a transcription.
Invoke
ableton midi transcribe <AUDIO> [--out <OUT.mid>] --json
Returns {output: "<path>.mid"}. Default output is alongside the input as
<stem>_basic_pitch.mid.
Tuned defaults (from the source project's sax transcription): onset_threshold 0.5, frame_threshold 0.3, minimum note 58 ms, melodia_trick on. These suit a monophonic lead; for denser material, mention it and we can expose overrides.
First run installs basic-pitch/TensorFlow (can take a minute). To sanity-check a
transcription, follow up with the midi-compare skill against a reference MIDI.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.