mcp-audio is an AIO-2030 compliant MCP plugin that performs voice-to-text transcription using the Audio speech recognition API.

It exposes the identify_voice method via both multipart/form-data and base64 formats, supports the AIO tools.call protocol, and returns JSON-RPC structured outputs.

• Fully AIO-compliant MCP plugin (/tools.call, /help)
• Converts .wav/.mp3 audio files to transcripts using SiliconFlow
• API key managed securely via .env file
• Docker-compatible and minimal dependencies
• Registration-ready for AIO endpoint registry

git clone [email protected]:AIO-2030/mcp-audio.git
cd mcp-audio
python -m venv venv && source venv/bin/activate
pip install -r requirements.txt

cp .env.example .env

Set your audio URL and API key:

AUDIO_URL=https--xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
API_KEY=sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

python src/mcp_server.py

docker build -t mcp-audio .
docker run --env-file .env -p 8080:8080 mcp-audio

Upload audio file directly. Response:

{
  "transcript": "hello world",
  "confidence": 0.91,
  "audio_hash": "a1b2c3..."
}

JSON-RPC format with base64-encoded audio. Response:

{
  "method": "tools.call",
  "params": {
    "method": "identify_voice",
    "inputs": [
      {
        "type": "audio",
        "value": "<base64-audio>"
      }
    ]
  }
}

Auto-serves contents of mcp_audio_registration.json. Used by Queen AI for MCP discovery and service indexing.

python test/test_audio_base64.py

python health_check.py

./register_mcp.sh

Requires jq, dfx, and a running endpoint_registry canister.