@audio/vocals

Try it in the browser: Vocal remover. Runs on this package, nothing is uploaded.

Center-channel vocal isolation / removal — the SoX oops complement, M/S based.

import { isolate, remove } from '@audio/vocals'

isolate([L, R])   // keep center-panned content (typically the vocal), discard sides
remove([L, R])    // keep sides, discard center — karaoke

Built on @audio/spatial-midside's encode/decode; mono passes through unchanged. ML stem separation is out of scope here (see @audio/mir deferred tier).