@audio/speech-vocoder

import vocoder from '@audio/speech-vocoder'

Channel vocoder — transfers the spectral envelope of one sound onto the pitched content of another.

Note: takes two separate buffers, always returns a new Float64Array (does not modify either input in-place).

Principle: analyze modulator into N log-spaced bandpass bands → extract envelope per band (asymmetric attack/release follower) → multiply with the same-band-filtered carrier → sum
Implementation: N parallel RBJ bandpass filters (Q=5, fixed) on both signals; envelope follower per modulator band (5 ms attack, 50 ms release)
Band count: 8 = robotic effect; 16 (default) = classic vocoder sound; 32+ = more speech intelligibility

// carrier: pitched source (sawtooth, buzz, noise...)
// modulator: signal whose spectral shape to impose (voice, instrument...)
let output = vocoder(carrier, modulator, { bands: 16, fs: 44100 })

Rebuilds the filter bank only when bands/fmin/fmax/fs actually change — safe to reuse the same params object across calls to carry filter and envelope state forward.

Inventor: Homer Dudley (1939), Bell Labs.

Use when: voice effects, talkbox simulation, cross-synthesis, spectral morphing.

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

MIT © audiojs