Utterance

All notable changes to this project will be documented here.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

[0.0.1] — 2025-02-18

Added

  • Core Utterance class with event-driven API (speechStart, pause, turnEnd, interrupt)
  • TurnDetector state machine for turn-taking logic with configurable sensitivity and pause tolerance
  • AudioCapture module using the Web Audio API (microphone access, real-time audio processing)
  • FeatureExtractor with energy computation and stubs for MFCCs, pitch, and speech rate
  • ONNXModel module for ONNX Runtime Web inference
  • EnergyVAD baseline classifier using RMS energy thresholds as fallback when no ONNX model is loaded
  • Dual-format build: ESM (dist/index.js) and CommonJS (dist/index.cjs) with TypeScript declarations
  • Shared type definitions in src/types.ts
  • Documentation site at utterance.dev with quick start, API reference, and architecture guides
  • Interactive playground at /playground for real-time event testing
  • 11 passing tests for the detector and feature extractor

Contributing

How to contribute to Utterance.

On this page

[Unreleased][0.0.1] — 2025-02-18Added