fish-speech-mcp

What is fish-speech-mcp?

fish-speech-mcp is an MCP server for text-to-speech synthesis (TTS) using FishSpeech. It integrates with MCP-compatible LLMs such as Dive and is intended for users who want to convert text to speech and save voice references for cloning.

How to use fish-speech-mcp?

Install via Dive Desktop by adding a new MCP server with the provided JSON configuration, or start the server manually with npx @demon24ru/fish-speech-mcp. Configure the Optivus server URL using the environment variable MCP_FISH_SPEECH_SERVER_URL. Invoke the tools text_to_speech (text, optional reference ID) and save_voice_reference (reference audio path and text).

Key features of fish-speech-mcp

• Text-to-speech conversion using FishSpeech
• Save voice references for future voice cloning
• MCP integration with Dive and other MCP-compatible LLMs
• Communication with Optivus server via Socket.IO
• Automatic connection, reconnection, and error handling

Use cases of fish-speech-mcp

• LLMs generating spoken responses from text input
• Cloning voices by saving a reference audio and corresponding text
• Integrating TTS capabilities into MCP-compatible applications

FAQ from fish-speech-mcp

What runtime does fish-speech-mcp require?

Node.js 20+ and an MCP-compatible LLM service.

How does fish-speech-mcp communicate with the TTS backend?

It connects to an Optivus server using Socket.IO, sending requests and receiving responses via the message event.

Where are voice references stored?

Voice references are stored in a directory from the Optivus server, each in a subdirectory named with a unique ID.

What environment variable configures the server?

MCP_FISH_SPEECH_SERVER_URL (default: http://localhost:5000).

How do I install fish-speech-mcp in Dive Desktop?

Add a new MCP server with the JSON configuration using npx -y @demon24ru/fish-speech-mcp as the command.