hass-speaker-recognition 1.0.7

Speaker recognition for Home Assistant using Resemblyzer

Speaker Recognition for Home Assistant

Identify speakers by their voice using machine learning. This project provides a complete speaker recognition solution for Home Assistant, including a REST API service, Python client library, custom integration, and Home Assistant addon.

✨ Features

• 🎤 Voice-based speaker identification using neural embeddings
• 🏠 Native Home Assistant integration with STT and conversation agents
• 🐳 Easy deployment via Home Assistant addon or standalone Docker
• 🔌 REST API for flexible integration with any platform
• 📦 Python client library for programmatic access
• 🎯 High accuracy powered by Resemblyzer voice embeddings
• ⚡ Fast recognition with cached embeddings
• 🔧 Configurable via UI or YAML

📋 Table of Contents

• Installation
  • Home Assistant Addon
  • Python Package
  • Docker
• Usage
  • Training
  • Recognition
  • Home Assistant Integration
• API Documentation
• Configuration
• Development
• Contributing
• License

🚀 Installation

Home Assistant Addon

The easiest way to use speaker recognition in Home Assistant:

1. Add this repository to your Home Assistant addon store
2. Install the Speaker Recognition addon
3. Configure the addon settings:
   • Host: 0.0.0.0 (default)
   • Port: 8099 (default)
   • Embeddings Directory: /share/speaker_recognition/embeddings
   • Log Level: info
4. Start the addon
5. Install the Speaker Recognition integration via the UI

Python Package

Install the client-only package (no ML dependencies):

pip install speaker-recognition

Install with server capabilities (requires Python <3.10):

pip install speaker-recognition[server]

Docker

Run the standalone service:

docker run -d \
  -p 8099:8099 \
  -v ./embeddings:/app/embeddings \
  ghcr.io/eulemitkeule/speaker-recognition:latest

📖 Usage

Training

Train the system with voice samples for each speaker:

Using Python Client

from speaker_recognition import SpeakerRecognitionClient
from speaker_recognition.models import TrainingRequest, VoiceSample, AudioInput

async with SpeakerRecognitionClient("http://localhost:8099") as client:
    training = await client.train(
        TrainingRequest(
            voice_samples=[
                VoiceSample(
                    user="Alice",
                    audio_input=AudioInput(
                        audio_data="<base64-encoded-audio>",
                        sample_rate=16000
                    )
                ),
                VoiceSample(
                    user="Bob",
                    audio_input=AudioInput(
                        audio_data="<base64-encoded-audio>",
                        sample_rate=16000
                    )
                )
            ]
        )
    )
    print(f"Trained {training.speakers_count} speakers")

Using REST API

curl -X POST http://localhost:8099/train \
  -H "Content-Type: application/json" \
  -d '{
    "voice_samples": [
      {
        "user": "Alice",
        "audio_input": {
          "audio_data": "<base64-audio>",
          "sample_rate": 16000
        }
      }
    ]
  }'

Recognition

Identify a speaker from audio:

Using Python Client

from speaker_recognition import SpeakerRecognitionClient
from speaker_recognition.models import RecognitionRequest, AudioInput

async with SpeakerRecognitionClient("http://localhost:8099") as client:
    result = await client.recognize(
        RecognitionRequest(
            audio_input=AudioInput(
                audio_data="<base64-encoded-audio>",
                sample_rate=16000
            )
        )
    )
    print(f"Speaker: {result.speaker} (confidence: {result.confidence:.2%})")

Home Assistant Integration

Once the integration is configured:

1. Configure the backend in the main integration entry
2. Map voices to users in the integration settings
3. Add STT entity as a sub-entry for speech-to-text with speaker ID
4. Add Conversation Agent as a sub-entry for voice commands with speaker context

The integration will automatically identify speakers and make the information available to your automations.

🔌 API Documentation

Endpoints

GET /health

Health check endpoint.

Response:

{
  "status": "healthy"
}

POST /train

Train the model with voice samples.

Request:

{
  "voice_samples": [
    {
      "user": "string",
      "audio_input": {
        "audio_data": "base64-string",
        "sample_rate": 16000
      }
    }
  ]
}

Response:

{
  "speakers_count": 2,
  "message": "Training completed successfully"
}

POST /recognize

Recognize a speaker from audio.

Request:

{
  "audio_input": {
    "audio_data": "base64-string",
    "sample_rate": 16000
  }
}

Response:

{
  "speaker": "Alice",
  "confidence": 0.95
}

⚙️ Configuration

Addon Configuration

host: "0.0.0.0"
port: 8099
log_level: "info"
access_log: true
embeddings_dir: "/share/speaker_recognition/embeddings"

Environment Variables

• HOST: Server host (default: 0.0.0.0)
• PORT: Server port (default: 8099)
• LOG_LEVEL: Logging level (default: info)
• ACCESS_LOG: Enable access logs (default: true)
• EMBEDDINGS_DIR: Directory for storing embeddings (default: ./embeddings)

🛠️ Development

Prerequisites

• Python 3.9 (for server development)
• Python 3.8+ (for client-only development)
• uv package manager

Setup

# Clone the repository
git clone https://github.com/eulemitkeule/speaker-recognition.git
cd speaker-recognition

# Install dependencies
uv sync --all-groups

# Run tests
uv run pytest tests/ -v

# Run linting
uv run ruff check .

# Run type checking
uv run mypy --strict speaker_recognition

Running Locally

# Start the server
uv run python -m speaker_recognition

# Or with custom options
uv run python -m speaker_recognition --host 0.0.0.0 --port 8099

Project Structure

speaker-recognition/
├── speaker_recognition/          # Main package
│   ├── api.py                   # FastAPI application
│   ├── client.py                # HTTP client
│   ├── models.py                # Pydantic models
│   └── recognizer.py            # Recognition logic
├── custom_components/           # Home Assistant integration
│   └── speaker_recognition/
├── speaker_recognition_addon/   # Home Assistant addon
├── tests/                       # Test suite
└── example_data/               # Example audio files

🤝 Contributing

Contributions are welcome! Please follow these steps:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Make your changes
4. Run tests and linting
5. Commit your changes (git commit -m 'Add amazing feature')
6. Push to the branch (git push origin feature/amazing-feature)
7. Open a Pull Request

Code Quality

• Follow PEP 8 style guidelines
• Use descriptive variable and function names
• Add type annotations
• Write tests for new features
• Keep methods focused and concise

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Resemblyzer - Neural voice embeddings
• Home Assistant - Home automation platform
• FastAPI - Modern web framework

📞 Support

• 🐛 Report bugs
• 💡 Request features
• 📖 Documentation

---

Made with ❤️ for the Home Assistant community