Real-time voice changer API for developers
The engine behind Vaux is available as an API. Stream microphone audio in and get the same words back in a different voice, live. Built for apps that swap faces or characters on video and need the voice to match.
Realtime WebSocket
Audio in, converted voice out, plus transcripts and events. About half a second to first audio.
Secure by default
API keys stay on your server; clients get short-lived, single-session tokens with server-enforced time caps.
MediaStream output
Merge the voice track with any face or character video model in the browser.
Usage you can bill
Per-session usage records and a usage API for passing costs to your users.
How it works
- 1
Activate the free API Sandbox and create a key.
- 2
Mint a client token on your server with POST /v1/tokens.
- 3
Connect with the JavaScript SDK and stream audio.
Need help with setup? Follow the step-by-step guide for your device.
Questions
Is there a real-time voice conversion API?
Yes. Vaux offers a WebSocket API and JavaScript SDK for live speech-to-speech, plus REST endpoints for text-to-speech, voices, voice cloning, sounds and usage.
Can I pair it with a face-swap or avatar model?
Yes. The SDK returns a standard MediaStream audio track you can combine with any video stream, so the voice matches the face or character.
How much does the API cost?
There is a free Sandbox plan, and paid API plans start at $49 per month. See the Developers page for limits and pricing.