FastDiarize 🎧 – FastAPI Speaker Diarization API

🚀 FastDiarize is a lightweight and efficient speaker diarization API powered by FastAPI and Pyannote.audio. It enables automatic speaker segmentation in audio files using AI.

---

🔥 Quick Start with Docker (Recommended)

1️⃣ Run the API in seconds with Docker (with GPU support):

docker run -d --gpus all -p 8000:8000 -e HUGGINGFACE_TOKEN="your_huggingface_token" lumymedia/fastdiarize:latest

💠 No installation needed! The API is instantly available at http://localhost:8000.

---

2️⃣ Analyze an audio file

Send a POST request to the /analyze endpoint with a publicly accessible MP3 URL:

curl -X 'POST' 'http://localhost:8000/analyze' \
     -H 'Content-Type: application/json' \
     -d '{"url": "https://example.com/audio.mp3"}'

---

3️⃣ Example API Response

The API returns timestamps for each speaker in the following format:

{
  "request_id": "123e4567-e89b-12d3-a456-426614174000",
  "segments": [
    {"speaker": "SPEAKER_1", "start": 0.00, "end": 3.50},
    {"speaker": "SPEAKER_2", "start": 3.51, "end": 7.80}
  ]
}

• request_id: A unique ID for the request.
• segments: A list of speaker segments with start and end times in seconds.

---

📝 Prerequisites

Before using FastDiarize, ensure you have the following:

• Hugging Face API token: Required to load the Pyannote.audio pipeline. Get your token here.
• Docker (optional but recommended): For easy deployment.
• Python 3.8+: If running manually.

---

⚡ Notes & Considerations

FastDiarize is designed to be lightweight and easy to use, but here are a few things to keep in mind:

• The audio file must be publicly accessible via a URL.
• Currently, only MP3 files are supported.
• For best results, audio should be at least 1 second long.
• Longer files may take more time to process.

Want to help improve these features? Pull requests are welcome! 🚀

---

🚀 Performance

• The API automatically uses GPU acceleration if available.
• Processing time depends on the length of the audio file.
• For optimal performance, ensure your system has a compatible GPU.

---

❓ Common Errors

Here are some common errors and how to resolve them:

• 408 Timeout: The audio download took too long. Check the URL and try again.
• 400 Bad Request: The audio file is invalid, too short, or inaccessible.
• 500 Internal Server Error: An unexpected error occurred during analysis. Check the logs for details.

---

🛋️ Manual Installation (Optional)

If you prefer to run the API manually:

1️⃣ Clone the repository

git clone https://github.com/lumymedia/FastDiarize.git
cd FastDiarize

2️⃣ Install dependencies

pip install -r requirements.txt

3️⃣ Set your Hugging Face API token

Create a .env file and add:

HUGGINGFACE_TOKEN=your_huggingface_token

4️⃣ Run the API locally

uvicorn main:app --host 0.0.0.0 --port 8000

---

🐳 Deploy with Docker Compose

For a fully managed deployment, use Docker Compose:

docker-compose up -d

---

🎯 Contributing

Want to improve FastDiarize? Contributions are welcome!

1️⃣ Fork the repo & create a new branch
2️⃣ Make changes & commit
3️⃣ Open a Pull Request

---

🐟 License

FastDiarize is released under the MIT License.

🌟 If you find this project useful, give it a star on GitHub! 🚀