@audio/sinusoidal-track

Peak picking + partial tracking

npm install @audio/sinusoidal-track

import track from '@audio/sinusoidal-track'

Sinusoidal analysis: STFT spectral peaks with parabolic interpolation, linked frame-to-frame into partial trajectories by nearest-frequency continuation within a cents threshold (McAulay & Quatieri 1986). Produces the model consumed by sinusoidal-synth (additive resynthesis) and sinusoidal-residual (noise layer).

let model = track(data, { frameSize: 2048, hop: 512 })
// model.partials: [{ start, freqs: number[], amps: number[] }, ...]

| Param | Default | | |---|---|---| | fs | 44100 | Sample rate | | frameSize | 2048 | STFT frame | | hop | 512 | STFT hop | | maxPartials | 40 | Peaks kept per frame | | threshold | -60 | Peak floor, dB relative to frame max | | maxJump | 60 | Max cents jump for partial continuation |

Returns { partials, frames, hop, frameSize, fs } — partials only includes trajectories lasting ≥3 frames.

Use when: the analysis stage of the sinusoids+noise (SMS) pipeline — additive resynthesis, residual extraction, or trajectory-based processing (smoothing, transposition).

Part of @audio/sinusoidal — the sinusoidal family umbrella.

MIT © audiojs