VoiceLabsWhat is VoiceLabs?
VoiceLabs is a browser-based AI voice studio — text-to-speech, voice cloning, and transcription running on hosted GPUs, with accounts, billing, an MCP server, and a public API.
VoiceLabs is a browser-based AI voice studio from Devino Solutions. Text-to-speech and voice cloning are designed to run on VoiceLabs' GPU servers and stream back to a tab in your browser — so there is nothing to install, no drivers, and no local GPU required.
The product lives at voicelabs.now, and the app itself at app.voicelabs.now.
Built in the open
VoiceLabs is in active development and is built in the open. Live today: the public site, account sign-up, the VoiceLabs Pro pricing and billing pages (with a 7-day free trial), a monthly free-generation allowance for signed-in accounts, the VoiceLabs MCP server for AI agents, and the public /v1 developer API. Planned at launch: the full in-browser studio with seven TTS engines and instant voice cloning. Real, dated release notes are posted on the changelog as features ship.
What you can do with VoiceLabs
- Generate speech from text in one of your voices. Generation runs on VoiceLabs' GPUs and is asynchronous: you get a generation id, then poll for the finished audio.
- Keep voice profiles — cloned voices and presets, each with a language, an engine, and usage counts.
- Transcribe audio with VoiceLabs' hosted Whisper; transcripts are saved as captures on your account.
- Connect AI agents through the remote MCP server, with OAuth 2.1 consent and per-agent revocation.
- Build on the API — a versioned
/v1REST surface with API keys and an official TypeScript SDK.
Open-source roots
VoiceLabs is built on the open-source VoiceBox project, which is MIT licensed, and that attribution is permanent. The hosted VoiceLabs service itself — the GPU inference, the accounts, the billing — is a commercial product from Devino Solutions.
Where to go next
Getting started
Create an account, understand the free allowance, and start the 7-day Pro trial.
Voices and engines
Voice profiles, the planned seven-engine lineup, and voice cloning rules.
Connecting AI assistants
Use VoiceLabs from Claude Code or any MCP client over OAuth 2.1.
Developer API
API keys, the /v1 REST endpoints, and the official TypeScript SDK.