Every sound has a pitch. This traces the pitch of your voice — the natural melody hidden in how you talk — and notates it as real sheet music. Record or upload a clip; vowels and held sounds become notes, pauses and unvoiced sounds become rests.
How it works
Your recording is sliced into overlapping windows and each one is run through the ACF2+ autocorrelation algorithm to estimate its dominant pitch, which is converted to the nearest MIDI note; consecutive windows with the same pitch are merged into a single note, silent or unvoiced stretches become rests, and the resulting sequence is quantized to note durations and rendered as real notation with VexFlow.
FAQ
Why does a short sound or breath sometimes turn into silence instead of a note?
Segments shorter than about 60 milliseconds are treated as pitch-detection glitches and converted to rests rather than notes — real spoken syllables are usually longer than that, so this filters noise without discarding legitimate short notes.
Can it detect pitch in whispered or unvoiced speech?
No — the autocorrelation algorithm needs a periodic waveform with enough signal energy to find a pitch; whispering and unvoiced sounds like 's' or 'f' produce no measurable pitch and become rests.
Free, unlimited, no account. If this saved you a paid tool, a small thanks is appreciated.