Whisper-Run

Whisper-Run is a pip CLI tool for processing audio files using Whisper models with speaker diarization capabilities. The tool allows you to process audio files, select models for audio processing, and save the results in JSON format.

It uses the OpenAI-Whisper model implementation from OpenAI Whisper, based on the ctranslate2 library from faster-whisper, and pyannote's speaker-diarization-3.1. Check their documentation if needed.

Before You Start

You must confirm the licensing permissions of these two models:

• pyannote/speaker-diarization-3.1
• pyannote/segmentation-3.0

From Hugging Face, get your Hugging Face auth token. You can put the token in your env file or pass it to the CLI as --hf_auth_token.

Installation

To install Whisper-Run, run the following command:

pip install whisper-run

Usage

You can call Whisper-Run from the command line using the following syntax:

whisper-run --file_path=<file_path>

Example

To process an audio file using the CPU and a specific file path:

whisper-run --device=cpu --file_path=your_file_path

When you run the command, you'll be prompted to select a model for audio processing:

[?] Select a model for audio processing:
 > distil-large-v3
   distil-large-v2
   large-v3
   large-v2
   large
   medium
   small
   base
   tiny

Flags

• --device: Specify the device to use for processing (e.g., cpu or cuda).
• --file_path: Specify the path to the audio file you want to process.
• --hf_auth_token: Optional. Pass the Hugging Face Auth Token or set the HF_AUTH_TOKEN environment variable.
• --save: Optional. If set, the results will be saved to a JSON file.

Output

Results are printed to the terminal as a JSON object. If the --save flag is set, the results are also saved in the results directory as a JSON file. You can change the output format in the audio_processor.py file.

Programmatic Usage

You can also use Whisper-Run programmatically in your Python scripts. Below is a basic usage example demonstrating how to use the Whisper-Run library:

Example Script

from whisper_run import AudioProcessor

def main():
    processor = AudioProcessor(file_path="your_file_path",
                               device="cpu",
                               model_name="large-v3",
                               hf_auth_token="your_hf_token",
                               save=True)
    processor.process()

if __name__ == "__main__":
    main()

Contributing

Contributions are welcome! Please open an issue or submit a pull request on GitHub.

License

This project is licensed under the Apache 2.0 License.