mcp-fish-audio

This MCP server exposes Fish Audio TTS capabilities to your LLM workflows. It integrates Fish Audio's text-to-speech API with your model prompts, allowing you to generate natural-sounding speech and stream or save audio outputs. Two tools are available: fish_audio_tts, which turns text into speech using a chosen voice reference (by ID, name, or tag), and fish_audio_list_references, which lists all configured voice references and the default reference. Use the tool parameters to specify the source text, select voices, control streaming, and choose output formats like MP3, WAV, PCM, or Opus. The server prioritizes voice selection as reference_id > reference_name > reference_tag > default, so you can fine-tune which voice is used for each request.

Prerequisites:

• Node.js (LTS) and npm installed on your system
• Access to install npm packages or run npx commands

Installation steps:

1. Install or upgrade Node.js and npm if needed

   • On macOS/Linux: curl -fsSL https://get.nodejs.org/install.sh | bash
   • On Windows: use the Node.js installer from https://nodejs.org/

2. Run the MCP server directly with npx (from anywhere): npx @alanse/fish-audio-mcp-server

3. Or configure via MCP settings (example provided in mcp_config):

   • Create or update your MCP config file with the provided mcpServers entry
   • Ensure environment variables (FISH_API_KEY, etc.) are set in the hosting environment

4. (Optional) Install globally for convenience: npm install -g @alanse/fish-audio-mcp-server fish-audio --help

5. Start the server in your environment (if not using npx in the MCP manager): npx -y @alanse/fish-audio-mcp-server

Tips and caveats:

• Always provide your Fish Audio API key via FISH_API_KEY in the environment where the MCP server runs.
• If using multiple voices, consider configuring FISH_REFERENCES as a JSON array for easier management (or use the individual FISH_REFERENCE_* vars).
• The default output directory is ~/.fish-audio-mcp/audio_output; ensure the hosting user has write permissions to this path or override AUDIO_OUTPUT_DIR.
• For streaming scenarios, set FISH_STREAMING to true and tune FISH_LATENCY for your network conditions.
• If you encounter authentication errors, verify API key scope and that the key is active for Fish Audio TTS.
• The server supports multiple formats (mp3, wav, pcm, opus); ensure your client can handle the chosen format.