@audio/shift-lpc

LPC source-filter pitch shift — repitched excitation residual through the unmodified all-pole vocal-tract filter

npm install @audio/shift-lpc

import lpc from '@audio/shift-lpc'

Canonical LPC source-filter pitch shift (residual-excited linear prediction, RELP lineage). Per frame: fit the vocal-tract all-pole filter A(z) by the autocorrelation method (Levinson-Durbin), inverse-filter the signal down to its spectrally-flat excitation residual, repitch that residual with the delay line splicer, then resynthesize through the unmodified 1/A(z) run continuously with block-switched coefficients. The synthesis filter's poles never move — the classical speech-processing complement to formant's cepstral envelope preservation, built on a different mechanism (an explicit source-filter model instead of a magnitude-envelope correction).

lpc(audio, { ratio: 1.5 })
lpc(audio, { ratio: 1.5, order: 24, frameSize: 512 })

| Param | Default | | |---|---|---| | order | min(frameSize/16, round(2 + sr/1000)) | AR filter order (pole count) | | frameSize | 1024 | Analysis/synthesis frame length |

Preserves formant frequencies by construction — the filter that carries them is never touched, only its excitation is repitched. Destroys pure tones by design: on a single sinusoid the AR envelope IS the partial, so the filter locks onto it and pitch barely moves — the family's defining tradeoff, not a bug.

| f0 err | THD% | alias | attack corr | formant dist | phase coh | shift | |-------:|-----:|------:|------------:|-------------:|----------:|------:| | 228.33 | 0.8 | 2.274 | 0.987 | 1.382 | 0.975 | 1.811 |

f0 err 228.33 Hz and alias 2.274 — both worst in the collection, and both the same degeneracy: on the alias test's 14 kHz tone (pushed toward Nyquist), the AR fit locks onto that single partial and the high-order synthesis filter rings into sharp amplitude spikes rather than shifting cleanly. Formant dist 1.382 is higher (worse) than formant's 0.765 on this synthetic vowel too, across every order from 8 to 64 — the fitted envelope follows the fixture's individual harmonics as well as its three formants, where formant's deliberately-smoothed cepstral envelope doesn't. None of this is the intended material: real, less strictly periodic speech is what an unmodified synthesis filter is for.

Use when: Speech/vocals where the formants must not move with pitch, and the material isn't a bare sustained tone. Not for: Pure tones, synth pads, or anything where the "partial" and the "envelope" are the same thing.

Stream

let write = lpc({ ratio: 1.5 })
let out = write(inputBlock)
let tail = write()  // flush

lpc buffers the whole input; write(chunk) accumulates and the batch algorithm runs once on write() (flush, no argument), returning everything at once.

Data

Input is a Float32Array (mono) or an array of Float32Array channels ([left, right, ...]) — anything else throws TypeError. ratio also accepts a function t => ratio (seconds from stream start) or a Float32Array breakpoint envelope (resampled across the input via ratioDuration, default the input's own duration) for time-varying pitch.

Part of @audio/shift — the shift family umbrella. This README is generated from the umbrella docs.

MIT © audiojs