LLMs are not and likely will never be good at translating things as they cannot and likely never will understand context, nor keep up with the way languages both natural and constructed change over time. Please do not use them as you will likely just confuse viewers.

As for programs to do subtitling yourself: https://alternativeto.net/software/subtitle-edit/?license=opensource

EDIT: seems like Whisper might be your best bet, it does speech-to-text + translation via a CLI that can be installed with pip: https://github.com/openai/whisper

Huggingface has loads of models for local speech to text and translation, see https://huggingface.co/docs/transformers/model_doc/speech_to_text

There is probably a nicer UI to use them than python coding, but I don’t know one.