AI-learning-material-from-video

This repository contains Python application designed for automatically transcribing video into text, translating the text, and burning the subtitles into the video. The application utilizes the Whisper model for transcription and optionally integrates with the OpenAI GPT-4 API for translation.

Features

• Automatic Transcription: Utilizes Whisper's model to transcribe audio into text, complete with timestamps.
• Subtitle File Creation: Converts transcription results into the SRT subtitle format.
• Translation: Offers an option to translate subtitles using manual input or automated translation via the OpenAI GPT-4 API.
• Subtitle Burning: Embeds subtitles directly into the video, providing an option to burn both original and translated subtitles.
• Generate Learning Material from video (TODO)

Update

• Added Groq inference for Whisper

Groq inference

Groq provides free and fast inference within a quota. Check out https://console.groq.com/playground. All you need to do is signing up and create an API key.

Setup

1. Clone the repository:

   git clone https://github.com/tracywong117/AI-learning-material-from-video.git
   cd AI-learning-material-from-video

2. Install required Python packages:

   pip install -r requirements.txt

3. Obtain an API key from OpenAI if you intend to use automated translation via GPT-4. Set this key in your environment variables or directly in your script securely.

Usage

1. Transcription and Subtitling:

   • Modify the input_video_file path in the script to point to your video file.
   • Set require_transcript to True if transcription is needed; otherwise, it can load an existing transcript.
   • Set require_burn_subtitle to True if burning subtitle is needed.
   • Set use_Groq to True if you want to use Groq inference instead of local inference (Whisper)
   • Set groq_api_key to your Groq Api Key
   • Run the script to transcribe, translate (if using the API), and burn subtitles:

     python whisper-transcript-subtitling.py

2. Translation and Subtitling:

   • Ensure the transcript SRT file is named correctly and in the same directory as your script.
   • If using the GPT-4 API for translation, ensure translate_option is set to "api" and your API key is correctly configured.
   • Set require_burn_subtitle to True if burning subtitle is needed.
   • Run the script:

     python gpt-translate-subtitling.py

Notes

• Review the subtitles for accuracy and synchronization issues after using the AI.

Support

For issues, suggestions, or contributions, please open an issue or a pull request in this repository.