mcp-audio

This MCP server exposes audio processing capabilities that can be invoked by any MCP-compatible client (e.g., Claude Desktop). It provides tools to transcribe audio, split large audio files into manageable segments, summarize transcripts, and analyze multiple transcripts in parallel. Clients connect via MCP and call the available tools by name: transcribe_audio, split_audio, summarize_transcript, and multi_file_chat. The server relies on Groq's Whisper models for transcription and supports intelligent splitting using silence detection, with robust export format fallbacks (MP3 → AAC → WAV).

To use it, configure an MCP client to point to this server with the uv-based command shown in the installation guide. For Claude Desktop, add an MCP server entry that uses uv to run the server-module and passes your GROQ_API_KEY as an environment variable. Then you can issue natural-language prompts like “Please transcribe this meeting” or “Please split the following long recording into under 25MB segments,” and the client will route your request to transcribe_audio or split_audio accordingly. For multi-file workflows, you can feed multiple transcripts to multi_file_chat and ask for cross-file insights or themes.

Prerequisites:

• Python 3.13+ installed on your system
• uv package manager installed (https://docs.astral.sh/uv/)
• ffmpeg installed and available in your PATH
• Groq API key (for transcription and summarization)

Installation steps:

1. Clone the repository:

git clone <your-repo-url>
cd mcp-audio-server

2. Install and run with uv (development workflow):

uv sync
uv run python -m mcp_audio_server.server

3. Set up your Groq API key:

export GROQ_API_KEY="your-groq-api-key-here"

4. Verify the installation by starting the server and testing a basic command:

uv run python -m mcp_audio_server.server

Standalone testing/helpful commands:

• Start the MCP server:

uv run python -m mcp_audio_server.server

• Test with the example client (if available):

uv run python examples/basic_usage.py

Configure Claude Desktop or another MCP client to point to the server using the provided configuration template in the README (adjust paths and keys as needed).

Tips and caveats:

• Ensure GROQ_API_KEY is exported in the environment where the MCP server runs; some clients require embedding the key in the MCP config too.
• ffmpeg must be installed for broad audio format support and correct export formatting.
• The server implements a 25MB limit handling strategy; use split_audio to chunk large files before transcription.
• For multi-file analysis, prepare file_paths lists and use multi_file_chat to extract cross-file insights.
• If you encounter issues with the MCP connection, verify the server path in your MCP configuration and that uv is installed and accessible in the environment running the server.
• The configuration example uses a cwd placeholder; update it to the actual directory containing the mcp-audio-server project.