This program performs the following tasks:
- Searches for MP4 Files: It identifies and processes MP4 video files.
- Converts MP4 to MP3: Converts the video files to MP3 audio format.
- Splits Audio Based on VTT Contents: Splits the MP3 audio into chunks based on VTT (WebVTT) subtitle timings. Each chunk should be no longer than 30 seconds, as the Whisper model supports up to 30-second audio chunks.
- Translates Audio Chunks: Uses the Whisper model to transcribe each audio chunk.
- Combines Transcriptions: Merges the transcriptions from all chunks into a single text file.
The program is designed to handle the transcription of audio extracted from video files, ensuring that each segment adheres to the Whisper model's requirements for chunk length.
Requirements:
!pip install git+https://github.com/openai/whisper.git
!apt-get install ffmpeg