I decided to create this subtitle file using bash scripts (I included all three of the scripts I used along with a README file). Basically, the bash script splits the audio into five minute chunks and then uses whisper-cli to translate each chunk. Because I wanted to use the best model available, the CPU use was heavy and could fail at times. So I added a cooldown between each chunk along with a finish.sh script in case any chunk failed. For this to work you need:
1) A linux distribution (I use Linux Mint)
2) **ffmpeg** and **ffprobe**
3) **whisper.cpp** built with `whisper-cli`
4) **Bash**
5) **perl** (used by the joiner for timestamp math; easy to swap for Python if you prefer)
I have no intention of creating a Windows version of this. Maybe the people at Microsoft will do it for you (since they seem to do all kinds of other things with your computer whether you want them to or not)
Incredibly, I was able to use ChatGPT to create these scripts (although I did have to give it some prompting and had to have it rewrite the script a few times)