VoiceStudio APIs
The local desktop API you can use today, and the Cloud API, which is still a preview.
Documentation
Build with your own voice, on your own machine.
Guides for the desktop app, a local OpenAI-compatible API, and an MCP server your agents can talk to. Nothing here needs an account.
Quickstart
Start the app and make your first speech request against the local API.
Connect with MCP
Let Cursor, Claude Code or Codex speak, clone and transcribe in your voice.
Local API
OpenAI-compatible audio on port 3900, with every route documented.
Desktop guides
Screen-by-screen guides for cloning, design, dubbing and audiobooks.
Model library
Every model, its languages, and the engine and hardware that runs it.
Cloud API preview
The hosted contract: Jobs, Artifacts and Usage. A preview; nothing is live yet.
Source on GitHub
Read the code, file an issue, or follow a release.
OmniVoice Studio is now VoiceStudio
Same open-source project, new name. What changed.
VoiceStudio has two APIs. Use the one your code talks to; a route in one is not promised by the other.
| Detail | Local desktop API | Cloud API |
|---|---|---|
| Status | Available while the app runs | Not released |
| Base URL | http://localhost:3900 | None yet |
| Version | 0.5.0 | 1.0.0 preview |
| Auth | None on localhost | Developer credential, once released |
Local desktop API
OpenAI-compatible audio, plus voices, dubbing, models and settings. Runs on your machine.
Cloud API preview
Hosted voices, dubbing and jobs. A contract you can read; nothing is live.
API audiences
Which interfaces are public, and which are never customer surface.
Are they compatible?
Not yet. Both name some /v1/audio/* routes, but the same spelling is not a
compatibility promise. Follow the quickstart against the
desktop API, and read the Cloud guide as a preview only.