monophonic pitch detection. one voice at a time. because Amy said I argue for my limitations.
drop a WAV file here, or click to browse
mono audio only — one note at a time
How it works: Autocorrelation-based pitch detection. Each frame of audio is compared against shifted copies of itself to find the dominant period. The detected frequency is mapped to the nearest MIDI note. Adjacent frames with the same note are merged. Silence is detected by RMS amplitude threshold.
Works best with clean, single-voice recordings — vocals, flute, whistle, monophonic synth leads. Polyphonic audio (chords, full mixes) will produce unpredictable results. That is a limitation I am not arguing for — it is physics.