This script generates a transcript of a YouTube video using the whisper model. The transcript is outputted in various formats, along with the original video and audio files, in a folder named after the video.
To use this script, you need to install the following dependencies:
yt-dlpffmpegXcodexcode-select
On the first run, the script clones whisper.cpp, downloads the large model, and compiles whisper. Once the model is compiled, the script will use it to transcribe the video.
To test the script, simply run ./test.sh. This will run the script on a sample YouTube video and output the transcript in various formats.
To transcribe any YouTube video, run ./main.sh <url-to-youtube-video>. Replace <url-to-youtube-video> with the URL of the YouTube video you want to transcribe. The script will download the video, extract the audio, and use the whisper model to transcribe the audio. The transcript will be outputted in various formats, along with the original video and audio files, in a folder named after the video.