VoiceLabs MCP¶

Speech for assistants, in voices you own. VoiceLabs is a hosted remote MCP server that lets an AI app speak text in your own voices, clone a voice from a recording you provide, and transcribe audio. Seven tools sit behind permissions you approve, with OAuth 2.1 sign-in instead of pasted API keys.

Server type: Remote (Streamable HTTP, stateless)
Auth: OAuth 2.1
Endpoint: https://app.voicelabs.now/api/mcp
Tools: list_voice_profiles, list_captures, get_generation, speak, transcription and voice tools
Pricing: vendor pricing (voicelabs.now)
Category: Communication
Built by: VoiceLabs (voicelabs.now, official registry name now.voicelabs/voicelabs)

Why This Matters for Operators¶

Voice is the last mile for a lot of operator work: narrated walkthroughs, onboarding audio, customer-facing voice prompts, meeting transcription. Doing it today means separate TTS, cloning and transcription products with their own keys and quotas. VoiceLabs puts all three behind one endpoint with permission-scoped tools, so an assistant can only reach the voices and actions you approved.

The engines run on the vendor's GPUs on open-source models, so the operator gets hosted convenience without betting the audio stack on a single proprietary voice vendor.

Tools & Capabilities¶

Tool area	Purpose
Voice profiles	List the account's cloned and preset voices, with past generations
Voice capture	List recent captures and their transcripts
Generation polling	Poll a generation for status, then read its audio URL
Speech	Generate speech from text, transcribe audio, add preset voices

Each tool sits behind a permission the consent screen shows before approval, so an app sees only what its permissions cover.

Installation¶

claude mcp add voicelabs --transport http https://app.voicelabs.now/api/mcp

Sign in with OAuth 2.1; there is no API key to paste. Engine and setup details live at voicelabs.now.

Configuration¶

{
  "mcpServers": {
    "voicelabs": {
      "type": "http",
      "url": "https://app.voicelabs.now/api/mcp"
    }
  }
}

Business Relevance¶

Content teams produce narrated assets without leaving the assistant workflow
Course and training operators generate lesson voiceovers in consistent voices
Support teams transcribe calls for QA and ticket enrichment
Agencies clone a client's approved voice for recurring creative production

Integration with CorpusIQ¶

VoiceLabs covers audio; CorpusIQ covers the business context the audio refers to. An assistant can pull Q3 numbers through CorpusIQ connectors and voice a client recap in the operator's own cloned voice, or transcribe support calls through VoiceLabs and match them to Stripe and CRM records through CorpusIQ connectors for a full conversation-to-revenue picture.

Limitations¶

Cloning requires a recording you provide, and built-in preset voices cannot clone a real person
Speech engines are open-source models hosted by the vendor, so quality ceiling is model-bound
Tool set is intentionally small (seven tools), focused on speech rather than audio editing
Pricing is not disclosed on the directory listing

FAQ¶

Can it clone any voice?¶

It clones from a recording you provide, and the built-in preset voices cannot clone a real person. Voice cloning is bounded by the permission model and the recordings you supply.

How is access controlled?¶

OAuth 2.1 sign-in with per-tool permissions: the consent screen shows each permission before you approve, and an app only sees the tools its permissions cover.

Where does the speech run?¶

On the vendor's GPUs using seven engines built on open-source models, listed at voicelabs.now/engines.