# Cartesia

> Cartesia is a paid MCP server for design and media. Voice AI from Cartesia: text-to-speech, transcription, voice cloning and pronunciation. It can run locally on your machine or as a hosted remote server.

- **Kind:** MCP server
- **Pricing:** Paid
- **Platforms:** API, CLI
- **Website:** https://docs.cartesia.ai/tools/ai/mcp
- **Repository:** https://github.com/cartesia-ai/cartesia-mcp
- **MCP Registry name:** io.github.cartesia-ai/cartesia-mcp
- **Categories:** design and media, AI models and LLM tooling
- **Listing updated:** 2026-11-14

## About

The server wraps Cartesia's Sonic text-to-speech and Ink speech-to-text models, so an agent can generate speech with a chosen voice, speed, volume and emotion, and transcribe audio in batch or streaming mode. Generated audio is saved by default and returned as a file id plus a download link valid for 24 hours.

Beyond synthesis, it covers voice management: listing and fetching voices, cloning from a sample, renaming or deleting clones, localising a voice into another language or dialect, and adding catalogue accents such as British to an instant clone. Pronunciation dictionaries can be created, updated and attached to speech requests, and an optional admin key unlocks credit usage reports for org admins.

There are two ways to run it. The hosted endpoint at mcp.cartesia.ai signs you in through the playground, and files there are referenced by file id. The local uvx package needs a Cartesia API key and writes audio to a directory you choose. The docs price text-to-speech at roughly one credit per character and voice agent calls at $0.03 per minute.

## Features

- **Text to speech:** Converts text to audio with voice, speed, volume, emotion and pronunciation dictionary options.
- **Speech to text:** Transcribes audio in batch or streaming mode.
- **Voice cloning:** Clones, updates and deletes voices from an audio sample.
- **Voice localisation:** Adapts a voice to another language or dialect and adds catalogue accents.
- **Pronunciation dictionaries:** Creates and manages dictionaries used in synthesis.
- **Hosted or local:** Use the hosted endpoint with sign-in, or run cartesia-mcp locally with uvx.

## Good for

- Generate narration for a script and download the audio
- Transcribe recorded calls or meetings
- Clone a voice and localise it into another language
- Fix mispronounced brand names with a dictionary
- Check credit usage for an organisation

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http cartesia-mcp https://mcp.cartesia.ai/mcp
```

### Claude Desktop (Settings → Connectors → Add custom connector)

```sh
https://mcp.cartesia.ai/mcp
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "cartesia-mcp": {
      "url": "https://mcp.cartesia.ai/mcp"
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "cartesia-mcp": {
      "type": "http",
      "url": "https://mcp.cartesia.ai/mcp"
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "cartesia-mcp": {
      "serverUrl": "https://mcp.cartesia.ai/mcp"
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "cartesia-mcp": {
      "type": "http",
      "url": "https://mcp.cartesia.ai/mcp"
    }
  }
}
```

## Tools

- `text_to_speech`: Converts text to audio and returns a file id and a 24-hour download URL by default.
- `speech_to_text`: Transcribes an audio file in batch or stream mode.
- `list_voices`: Lists voices, filterable by language, search and gender.
- `get_voice`: Fetches metadata for a voice by ID.
- `clone_voice`: Clones a voice from an audio sample.
- `update_voice`: Updates a cloned voice's name or description.
- `delete_voice`: Deletes a cloned voice.
- `localize_voice`: Adapts a voice to another language or dialect.
- `list_accents`: Lists catalogue accent IDs and metadata.
- `list_models`: Lists callable Sonic and Ink model IDs.
- `add_voice_accents`: Adds catalogue accents to an instant voice clone.
- `delete_voice_accent`: Removes a catalogue accent from an instant voice clone.
- `list_pronunciation_dicts`: Lists pronunciation dictionaries.
- `create_pronunciation_dict`: Creates a pronunciation dictionary.
- `get_pronunciation_dict`: Gets a pronunciation dictionary by ID.
- `update_pronunciation_dict`: Updates a pronunciation dictionary.
- `delete_pronunciation_dict`: Deletes a pronunciation dictionary.
- `download_file`: Fetches a cloud file by file ID and returns a fresh download URL.
- `get_credit_usage`: Shows credit usage over time; org admins only.

## FAQ

### Which models does it use?

Sonic for text-to-speech and Ink for speech-to-text; list_models returns the callable ids.

### Do I need an API key?

The hosted server uses a sign-in; the local uvx package needs CARTESIA_API_KEY, plus an optional admin key for credit usage.

### How long do download links last?

Links returned by text_to_speech are time-limited to 24 hours; download_file issues a fresh one.

### How is usage billed?

In credits: about one per character for text-to-speech, with voice agent calls at $0.03 per minute.

### Is Cartesia free?

No. Cartesia is a paid product.

### How do I add Cartesia to Claude Code?

Run this command in your terminal: claude mcp add --transport http cartesia-mcp https://mcp.cartesia.ai/mcp

### What are the best alternatives to Cartesia?

The closest alternatives to Cartesia are HyperFrames by HeyGen, ffmpeg-skill, Claude Code Ultimate Guide, AvatarSay — Fictional Character & Talking Video MCP and ChartAnt.

## Alternatives

- [HyperFrames by HeyGen](https://ai-powered.com/mcp/hyperframes-by-heygen): HyperFrames by HeyGen is a freemium MCP server for design and media.
- [ffmpeg-skill](https://ai-powered.com/mcp/ffmpeg-skill): ffmpeg-skill is an open-source MCP server for design and media.
- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [AvatarSay — Fictional Character & Talking Video MCP](https://ai-powered.com/mcp/avatarsay-fictional-character-and-talking-video-mcp): AvatarSay — Fictional Character & Talking Video MCP is a freemium MCP server for design and media.
- [ChartAnt](https://ai-powered.com/mcp/chartant): ChartAnt is a freemium MCP server for design and media.
- [VfxSmith](https://ai-powered.com/mcp/vfxsmith): VfxSmith is a freemium MCP server for design and media.
- [CalderFlow](https://ai-powered.com/mcp/calderflow): CalderFlow is a free MCP server for design and media.
- [Hourlore](https://ai-powered.com/mcp/hourlore): Hourlore is a freemium MCP server for design and media.

---
Source: https://ai-powered.com/mcp/cartesia
