New Feature: Support for Groq API in Whisper Transcription Tool

[matheushmfr]

Hello everyone,

I’ve opened a pull request to introduce support for the Groq API the ts audio transcription tool. Idea here is that we can use groq as a cheaper option: currently $0.03/hour transcribed vs Open AI's $0.006/minute transcribed (groq is 12x cheaper), plus also comes with a nice free tier.

Key Features Added:

1. Groq API Integration:

   • The Whisper class now supports transcription using the Groq API, in addition to OpenAI. This provides users with more flexibility and options for their transcription needs.

2. Model Fetching and Listing:

   • Implemented methods to fetch available models from both OpenAI and Groq APIs.
   • Added functionality to list available models, allowing users to easily see which models they can use.

3. Enhanced Argument Handling:

   • Improved the command-line interface to allow users to specify models and list available models.
   • Enhanced error handling to provide informative messages when API keys are missing or models are unavailable.

4. Temporary File Handling for Audio Segments:

   • Refactored audio processing to use temporary files, ensuring deletion of previous leftover audio segment files.

How to Use the New Features:

1. List Available Models:

   ts --listmodels

2. Specify a Model for Transcription:

   ts <audio_file_path> --model <model_name>

Link to the Pull Request:

Link to the pull request here. Please let me know if there are any questions or further improvements needed.

Hope these new features will be beneficial to the community.