Sinusoids + noise + transients modeling (McAulay-Quatieri 1986, Serra-Smith SMS). All planned.
| Package | What |
|---|---|
@audio/sinusoidal-track |
peak picking + partial tracking |
@audio/sinusoidal-synth |
additive resynthesis, trajectory smoothing |
@audio/sinusoidal-residual |
residual noise layer |
The De-Slop ambitious path ("gaussian splats for sound") — smooth partial trajectories to remove AI-codec pitch jitter and formant wobble. @audio/shift-sms / @audio/stretch-sms are the first consumers; speech-world (WORLD vocoder) builds on the same decomposition.
