Transcribe Srt

Dependencies

This skill depends on having whisper-cli available in your $PATH. It may also depend on ffmpeg being available if the input file isn't directly processable by whisper-cli. Finally, we will depend on python or python3 being available to execute the script wsp2srt.py as a part of this skill.

Instructions

If the input file type is not processable by whisper-cli then first call ffmpeg -i "{INPUT FILE}" "{INPUT FILE}.wav". This will output a .wav file that whisper-cli can understand.

Then call whisper-cli -m ~/whisper.cpp/models/ggml-large-v3-turbo.bin -vm ~/whisper.cpp/models/ggml-silero-v6.2.0.bin -f "{INPUT FILE}.wav" > almost_srt.txt. If you think the input file is mostly spoken in a language other than English, then add -l {language} to the command.

Finally, call python wsp2srt.py almost_srt.txt "{INPUT FILE}.srt" to save the SRT subtitles to a file that is similarly named as the input.