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

Articulatory vocal-tract synthesis — Kelly-Lochbaum waveguide (Pink Trombone class)

Voice synthesis — umbrella for @audio/voice-* atoms (vocal tract, voder, glottal source)

Teach your AI to play piano and guitar, sing, and now LISTEN. 53 MCP tools, 108 annotated songs (14 with redistributable MIDI shipped; the rest fetched from source), 6 engines, a browser cockpit, and an audio inspector that measures a recording: onsets, p

Portable WASM embedding generation with SIMD and parallel workers - run text embeddings in browsers, Cloudflare Workers, Deno, and Node.js

A package for easily working with US and state metadata

Center-channel vocal isolation & removal

Sonic library, compiled to WASM. Intended for use with Echogarden.

A modularized version of the Pink Trombone voice synthesizer

React component and hook for speech recognition, with voice commands and pluggable STT engines

An object-oriented language for parsing and pattern matching

Functional wrapper around the SpeechRecognition Web API

Provider-neutral text, interruption, telemetry, and testing reliability for TypeScript voice pipelines

Brain-Computer Interface (BCI) neural signal transcription addon for qvac, powered by whisper.cpp

RIO voice server, Vertical Service Codes (VSC)

Parallel formant filter bank

6 racks built only from Live's own EQ Eight, Saturator, Utility and Limiter: 33 devices and 112 settings, each one with the reason it is where it is. The streaming master rack sits at -1 dBTP, the

A set of fast primitives for entirely client-side formant detection.

Handles bulk geocoding for United States addresses with the U.S. Census Bureau's free batch geocoding API.

6 mixing racks, 21 stock Live devices, 85 parameters - and the reason for every single one. Master ceiling set to -1.0 dB because Live's Limiter reads sample peaks, not true peaks: the overshoot a str