A browser-based tool that converts singing voice into printable musical notation, with adjustable parameters for pitch tolerance, noise gate, and note splitting sensitivity.
# Voice → Notation
Source: [https://om-intelligence.ch/projects/vocal-notation/vocal-notation.html](https://om-intelligence.ch/projects/vocal-notation/vocal-notation.html)
Tempo90 BPMGridBeats per bar
Pitch tolerance±50 ¢Noise gate0\.003Click volume40%Re\-attack sensitivity0\.45
**Re\-attack sensitivity**decides whether two same\-pitch cells are one held note or two repeated notes\. It listens for the dip in volume between them: a note is split when the quietest moment at the boundary falls below this fraction of the note's own loudness\. Raise it to split more eagerly, drop it to 0 to never split — everything then becomes one long tied note, as before\. Sing repeated notes on separate syllables \("ta\-ta"\) and it works; sing them legato on one vowel and nothing in the signal distinguishes them from a held note\.
Use headphones — otherwise the click bleeds into the mic\. The noise gate ignores anything quieter than the threshold; the tolerance rejects pitches further than that from a real semitone, which cuts slides and scoops\.
Kyutai releases MuScriptor, an open model for multi-instrument transcription that converts any music recording into MIDI notes, similar to OpenAI Whisper for speech.