# Cartesia MCP

> Official Cartesia (Sonic) server: low-latency TTS, voice cloning, infill, localization.

- **Category:** Voice & Dialogue
- **Vendor:** Cartesia (official)
- **Pricing:** freemium
- **Verified:** 2026-06-11

## What it does

Cartesia's official MCP server exposes its Sonic voice stack: TTS, speech-to-text, voice cloning, voice changing, and notably audio infill (generate speech between two existing clips) and voice localization into other languages. Strong for ADR-style dialogue patching and localizing a character's voice across language versions.

## Connect

- Claude Code: `claude mcp add cartesia -e CARTESIA_API_KEY=YOUR_KEY -- uvx cartesia-mcp`
- Local: `uvx cartesia-mcp`

**Auth:** api_key (env `CARTESIA_API_KEY`) — https://play.cartesia.ai/keys

## Capabilities

`tts` · `stt` · `voice-cloning` · `voice-changing` · `infill` · `localization`

## Sample tools

`text_to_speech` · `speech_to_text` · `list_voices` · `clone_voice` · `voice_change` · `localize_voice` · `infill` · `get_credit_usage`

## Field notes

Officially supported, on PyPI as cartesia-mcp; ~16 tools. Optional CARTESIA_ADMIN_API_KEY enables account-management tools; OUTPUT_DIRECTORY controls where audio is saved.

## Links

- Site: https://www.cartesia.ai
- Docs: https://docs.cartesia.ai/tools/ai/mcp
- Repo: https://github.com/cartesia-ai/cartesia-mcp

---

Structured data: https://mcp.film/api/mcps/cartesia.json · Directory: https://mcp.film