@audio/tune

Pitch correction (Auto-Tune / Melodyne class).

| Package | What | |---|---| | @audio/tune-snap | scale/key correction: pYIN pitch → the notes meant on the scale → each centred on its note, slow drift flattened → one curve through tune-curve | | @audio/tune-midi | melody correction — YIN → per-guide-note median → PSOLA retune to a MIDI/note guide | | @audio/tune-curve | a drawn pitch curve: the voice's own glottal cycles re-spaced (TD-PSOLA): formants and consonants kept, streaming |

import { snap, tuneMidi } from '@audio/tune'

snap(data, { scale: 'major', root: 0 })      // to C major, vibrato kept
snap(data, { speed: 0 })                     // the hard-tuned effect
tuneMidi(data, { guide: parsedSmf.notes })   // retune to a reference melody
tuneCurve(data, { semitones: { t: [0, 1, 1.5], v: [0, 2, 0] } })   // along a curve, as drawn (Melodyne class)

tune-snap builds one pitch curve over the whole take and retunes along it with tune-curve, so notes are never cut out and crossfaded back over the dry voice. tune-midi still shifts each guide note apart with @audio/shift-psola and crossfades it back.