Is my audio uploaded anywhere?
No. Transcription runs entirely in your browser: audio is decoded locally and note detection runs in an on-device neural network inside a Web Worker. Your file never leaves your device.
SONGA TOOLS
MP3 to MIDI
Audio stays on this device. Transcription uses Spotify Basic Pitch.
Note transcription, not audio compression
MIDI stores note pitches, timing and duration, not the original voice or instrument sound. Clean, single-instrument recordings are better suited to transcription than full mixes.
How it works
- Choose a local audio file (MP3/WAV/OGG/FLAC, up to 20 MB and 120 seconds).
- Audio is decoded and resampled in the browser, then the Spotify Basic Pitch neural network detects notes inside a Web Worker.
- Tune sensitivity with the threshold slider, then download the .mid file when transcription finishes.
FAQ
Which formats and lengths are supported?
Any browser-decodable audio such as MP3, WAV, OGG or FLAC, up to 20 MB and between 0.1 and 120 seconds long.
What audio gives the best results?
Clean single-instrument or melodic recordings (piano, guitar, sung melody) work best. Full mixes, drums and noisy recordings produce stray notes — lower the detection threshold to catch more notes or raise it to reduce noise.
Can I use the MIDI file directly?
Yes. The generated .mid file imports straight into any DAW (FL Studio, Ableton, Logic, GarageBand and more) where you can edit notes and swap instruments.
Is it free?
Yes — transcription and MIDI download are free, with no account required.