# Clone the repository
git clone https://github.com/ebmarquez/audio-transcription-mcp.git
cd audio-transcription-mcp

# Create .env file with your HF token
echo "HF_TOKEN=hf_your_token_here" > .env

# Build and run with Docker Compose
cd docker
docker compose up -d

# Container is now running at http://localhost:8080/mcp

# CPU version
docker run --rm \
  -e HF_TOKEN="hf_your_token" \
  -v $(pwd)/input:/input:ro \
  -v $(pwd)/output:/output \
  -v $(pwd)/models:/root/.cache \
  -p 8080:8080 \
  audio-transcription-mcp:cpu

# GPU version (NVIDIA)
docker run --rm --gpus all \
  -e HF_TOKEN="hf_your_token" \
  -v $(pwd)/input:/input:ro \
  -v $(pwd)/output:/output \
  -v $(pwd)/models:/root/.cache \
  -p 8080:8080 \
  audio-transcription-mcp:gpu

# Clone and install
git clone https://github.com/ebmarquez/audio-transcription-mcp.git
cd audio-transcription-mcp
pip install -e .

# Set up environment
cp .env.example .env
# Edit .env and add your HF_TOKEN

# Run MCP server
python -m audio_transcription_mcp

{
  "mcpServers": {
    "audio-transcription": {
      "url": "http://localhost:8080/mcp",
      "transport": "streamable-http"
    }
  }
}

{
  "mcpServers": {
    "audio-transcription": {
      "command": "python",
      "args": ["-m", "audio_transcription_mcp"],
      "env": {
        "HF_TOKEN": "${HF_TOKEN}",
        "OUTPUT_DIR": "./transcriptions"
      }
    }
  }
}

transcribe_audio(
    file_path="/input/meeting.mp3",
    output_dir="/output",
    model_size="large-v3",
    include_timestamps=True,
    generate_summary=True
)

transcribe_directory(
    directory_path="/input",
    output_dir="/output",
    recursive=False
)

get_transcription_status(file_path="/input/meeting.mp3")

# Audio Transcription: meeting-recording.mp3

## Metadata
- **Source File**: meeting-recording.mp3
- **Duration**: 45:32
- **Speakers Detected**: 3
- **Transcription Date**: 2026-01-29
- **Model**: faster-whisper large-v3

---

## Transcript

### [00:00:00] **Speaker 1**
Good morning everyone. Let's get started with our weekly sync.

### [00:00:05] **Speaker 2**
Thanks for organizing this. I have a few updates on the project.

...

---

## Summary
[AI-generated summary placeholder]

## Key Points
- Point 1 extracted from conversation
- Point 2 extracted from conversation

## Action Items
- [ ] Action item 1 - Assigned to: Speaker 1
- [ ] Action item 2 - Assigned to: Speaker 2

audio-transcription-mcp/
├── docker/
│   ├── Dockerfile.cpu          # CPU container
│   ├── Dockerfile.gpu          # GPU container (NVIDIA)
│   ├── docker-compose.yml      # Development compose
│   ├── docker-compose.prod.yml # Production compose
│   └── entrypoint.sh           # Container startup
├── src/
│   └── audio_transcription_mcp/
│       ├── __init__.py
│       ├── __main__.py         # Entry point
│       ├── server.py           # MCP server
│       ├── config.py           # Configuration
│       ├── audio_processor.py  # File handling
│       ├── transcriber.py      # Faster-Whisper
│       ├── diarizer.py         # pyannote.audio
│       ├── segment_merger.py   # Align segments
│       └── markdown_generator.py
├── tests/
├── input/                      # Audio files (mount point)
├── output/                     # Transcriptions (mount point)
├── models/                     # Model cache (mount point)
├── .env.example
├── pyproject.toml
└── requirements.txt