Roadmap

• Core SDK with event-driven API (speechStart, pause, turnEnd, interrupt)
• EnergyVAD baseline classifier
• Published to npm as @utterance/core
• Documentation site and interactive playground
• Train and ship the semantic endpointing model (ONNX)
• ONNX Runtime Web inference integration
• React hook (useUtterance)
• CI/CD pipeline with automated npm publishing
• Framework adapters (Vue, Svelte)
• Voice SDK integration examples (LiveKit, Daily, Twilio)

Have a feature in mind? Open an issue or join the Discord.