CrisperWhisper Worker for RunPod

A serverless worker that provides high-quality speech transcription using CrisperWhisper on the RunPod platform.

Features

• Automatic speech transcription with CrisperWhisper
• Support for different languages
• RunPod serverless compatibility
• GPU acceleration

Input Parameters

Parameter	Type	Required	Default	Description
audio_file	string	Yes	N/A	URL to the audio file for transcription
language	string	No	null	ISO code of the language spoken in the audio (e.g., 'en', 'fr')
batch_size	int	No	16	Batch size for processing
temperature	float	No	0	Temperature to use for sampling (higher = more random)
debug	bool	No	false	Whether to print debug information

Usage Example

{
  "input": {
    "audio_file": "https://example.com/audio/sample.mp3",
    "language": "en",
    "debug": true
  }
}

Output Format

The service returns a JSON object structured as follows:

{
  "transcription": "The full transcribed text from the audio file.",
  "detected_language": "en"
}

Secrets Setup

This worker requires a Hugging Face token to download the CrisperWhisper model. Add your token as a RunPod secret named HF_TOKEN, which will be available in the container as RUNPOD_SECRET_HF_TOKEN.

Testing Locally

To test this worker locally:

1. Install Docker
2. Build the Docker image:

   docker build -t crisperwhisper-worker .
   

3. Run the container:

   docker run -e RUNPOD_SECRET_HF_TOKEN=your_hf_token -p 8000:8000 crisperwhisper-worker
   

RunPod Deployment

1. Create a new Serverless template
2. Connect your GitHub repository
3. Add your Hugging Face token as a secret
4. Deploy the template

License

This project is licensed under the Apache License, Version 2.0.

Acknowledgments

• This project utilizes code from WhisperX, licensed under the BSD-2-Clause license
• Special thanks to the RunPod team for the serverless platform

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.