PyAnnote Speaker Diarization RunPod Handler

This repository contains a RunPod handler for speaker diarization using PyAnnote. It takes audio input and returns diarization results with speaker segments and speaker embeddings.

Setup

Prerequisites

1. A Hugging Face account with access to the PyAnnote model
2. RunPod account with API access
3. Docker installed locally (for building the image)

Environment Variables

Set the following environment variable when deploying:

• HF_TOKEN: Your Hugging Face API token with access to the PyAnnote model

Building the Docker Image

docker build -t your-username/diarization-handler:latest .
docker push your-username/diarization-handler:latest

Deploying on RunPod

1. Go to your RunPod Serverless dashboard
2. Create a new endpoint using your Docker image
3. Set the required environment variables
4. Deploy your endpoint

API Usage

Send a POST request to your RunPod endpoint with the following structure:

{
  "input": {
    "audio_data": "<base64_encoded_audio>",
    "file_type": "wav"
  }
}

Parameters

• audio_data: Required. Base64-encoded audio file (wav, mp3, etc.)
• file_type: Optional. File format of the audio (default: "wav")

Response

{
  "diarization": [
    {
      "speaker": "SPEAKER_0",
      "start": 0.0,
      "end": 2.5
    },
    {
      "speaker": "SPEAKER_1",
      "start": 2.7,
      "end": 5.2
    }
  ],
  "embeddings_dict": {
    "SPEAKER_0": [0.1, 0.2, ...],
    "SPEAKER_1": [0.3, 0.4, ...]
  },
  "processing_time": 3.45
}

Local Testing

To test locally before deploying:

# Export your HF token
export HF_TOKEN="your_huggingface_token"

# Run the handler locally
python handler.py

You can then test it with:

python test.py

Example Code

import requests
import base64
import os

# Read and encode audio file
with open("audio.wav", "rb") as audio_file:
    audio_base64 = base64.b64encode(audio_file.read()).decode('utf-8')

# Set up headers with RunPod API key
headers = {
    'Content-Type': 'application/json',
    'Authorization': f'Bearer {os.getenv("RUNPOD_API_KEY")}'
}

# Prepare the request payload
json_input = {
    "input": {
        "audio_data": audio_base64,
        "file_type": "wav"
    }
}

# Send the request to your RunPod endpoint
response = requests.post('https://api.runpod.ai/v2/your-endpoint-id/runsync', 
                         headers=headers, 
                         json=json_input)

# Process the response
result = response.json()