Text to speech MCP server: give your AI assistant a voice
MCP (Model Context Protocol) lets an AI assistant use outside tools. The SuperClonica MCP server gives it a voice: the assistant finds a voice, voices your text and sends back a link to the audio file.
It works with the same API key and the same plan as the SuperClonica speech API, so a free key is enough to start.
What the server does
The server lives at https://superclonica.com/mcp and offers three tools. list_voices searches the stock voices and every catalog voice (characters, narrators, announcers) by name in any language, tags and description. speak voices a text with one voice and returns a link to an MP3, OGG or WAV file that works for an hour. get_usage tells the plan and how many seconds of audio are left.
Connect Claude Code
Create a free key on the speech API page, then run one command in a terminal: claude mcp add --transport http superclonica https://superclonica.com/mcp --header "Authorization: Bearer sc_YOUR_KEY". Replace sc_YOUR_KEY with your key. Claude Code lists the three tools in /mcp, and the assistant calls them when you ask for speech.
Connect Cursor or VS Code
In Cursor, add the server to ~/.cursor/mcp.json: {"mcpServers": {"superclonica": {"url": "https://superclonica.com/mcp", "headers": {"Authorization": "Bearer sc_YOUR_KEY"}}}}. In VS Code, the file is .vscode/mcp.json: {"servers": {"superclonica": {"type": "http", "url": "https://superclonica.com/mcp", "headers": {"Authorization": "Bearer sc_YOUR_KEY"}}}}. Any other MCP client that supports remote HTTP servers with a custom header works the same way: the address plus the Authorization header.
What to ask
Ask in plain words, for example: voice the sentence "Welcome to the dark side" in Darth Vader’s voice; find a calm female narrator and read this paragraph in English; how many seconds do I have left today. A voice can be named or given by its id; when several voices share a name, the assistant gets a short list to choose from. Clients that play audio can ask for the sound itself inside the answer.
Limits and plans
The MCP server is part of the speech API plans. A free key voices 3 minutes of audio a day with any catalog voice; paid plans start at $5 a month for up to 216 hours of audio. One call takes up to 20,000 characters, and when a limit is reached the assistant gets a plain message with the time to retry instead of a failure.