Cartesia MCP Server
Cartesia TTS and STT tools allow AI models to manage audio and voice cloning tasks.
Data last scanned today · Reviewed today
Overview
The Cartesia MCP server provides a direct interface for AI models to interact with the Cartesia API. It enables advanced audio operations including text-to-speech, speech-to-text transcription, and voice management capabilities like cloning and localization directly through the Model Context Protocol.
Our verdict
This server carries a high-risk profile with a 30/100 security score. It is community-maintained rather than official, lacks a defined license, and its security mechanisms—specifically authentication and read-only mode—have not been independently reviewed. It is intended for users comfortable with self-managed, unverified third-party software who require programmatic control over Cartesia audio services.
- Supports voice cloning, TTS, and STT transcriptions
- Requires zero direct dependencies for operation
- Active development with recent commits
- Clear documentation provided for integration steps
- Missing license creates legal usage ambiguity
- Not verified by official vendor maintainers
- Security controls for authentication remain unreviewed
Tools
| Tool | Description | Risk |
|---|---|---|
| text_to_speech | Convert text to audio; optional speed, volume, emotion, and pronunciation dict. Default save=true returns file_id and a 24h download_url. | medium |
| speech_to_text | Transcribe an audio file (mode=batch default, or mode=stream) | low |
| list_voices | List available voices (filter by language, search, gender, etc.) | low |
| get_voice | Fetch metadata for a voice by ID | low |
| clone_voice | Clone a voice from an audio sample | medium |
| update_voice | Update a cloned voice's name or description | medium |
| delete_voice | Delete a cloned voice | high |
| voice_change | Re-render audio with a different voice | medium |
| localize_voice | Adapt a voice to another language or dialect | medium |
| add_voice_accents | Add catalog accents to an instant voice clone (british, parisian, …) | medium |
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf | ||||
| ChatGPT |
Frequently asked questions
›What software do I need installed to run the Cartesia MCP server?
You need the uv package manager, which automatically handles Python 3.13+ installation and runs the server via uvx.
›Are there different API keys for standard and administrative features?
Yes, standard API keys are used for general tasks like TTS and STT, while an admin API key is required specifically for management tools like get_credit_usage.
›How can I change where the server saves generated audio files?
By default, files are saved in the working directory, but you can specify a custom path by adding an OUTPUT_DIRECTORY variable to your environment configuration.
›Does the server support local audio file processing?
Yes, tools like speech_to_text and voice_change accept paths to local audio files, which you should provide to your agent as full file paths.
›How does the server handle audio file downloads after text-to-speech conversion?
The text_to_speech tool returns a file ID and a download URL valid for 24 hours by default.
Changelog
- Fixcartesia-mcp: v0.22.0
2026-09-09
- Newcartesia-mcp: v0.21.0
2026-08-27
- Fixcartesia-mcp: v0.20.0
2026-08-20
- Newcartesia-mcp: v0.19.0
2026-08-14
- Fixcartesia-mcp: v0.18.2
2026-07-31
- Fixcartesia-mcp: v0.18.1
2026-07-25
- Newcartesia-mcp: v0.18.0
2026-07-25
- Fixcartesia-mcp: v0.17.1
2026-07-24
- Newcartesia-mcp: v0.17.0
2026-07-23
- Fixcartesia-mcp: v0.16.1
2026-07-15
Badge
Maintain this server? Add the live badge to your README.