From the node-webrtc-rust team
VoiceThere is built by the creators of the open-source @node-webrtc-rust/sdk npm package — develop on your laptop, deploy to our cloud with the same runtime.GitHub · npm
Early access
Voice AI platform for Node.js apps
Deploy voice agents without the WebRTC or session ops — we manage signaling, scaling, and infra. Free local STT and TTS on every worker; you ship agent logic in TypeScript. Join the waitlist for early access on voicethere.io.
Get early access
Why VoiceThere
We handle WebRTC, routing, and session lifecycle — you ship Node.js voice agents.
Deploy without infra stress
Ship an agent bundle with the CLI — we run signaling, TURN, and workers. No hand-rolled Kubernetes or WebRTC ops on your side.
Sessions orchestrated for you
Warm pools, assignment, teardown, and ICE/signaling routing — plus live debug attach to running calls. No DIY session state machines.
Pod-local STT by default
Sherpa-ONNX transcribes on the worker — audio stays on your pod for privacy, with faster turnaround and no per-minute STT API or extra network hop.
Pod-local TTS by default
Synthesize replies on-box out of the box — private, low-latency playback without shipping text to a cloud TTS API. Switch to a hosted voice when you need one.
Rust-native media core
Sub-10ms-class barge-in, VAD, and mixing in Rust — plus low-latency paths for pod-local STT and TTS. Node.js only for your business logic.
Bring your own AI vendors
Mix OpenAI, Deepgram, ElevenLabs, Google, Cartesia, or AssemblyAI per session when you want cloud STT, TTS, or LLMs.