Here is a breakdown of the files included

reduce2 - reducing audio quality to make it easier to process

Convert2 - converting raw audio data to text and then prompt response pairs

CorpusCounter - getting simple analytics on raw dataset

Final - where model is defined and finetuned

FinalTest - where the model is tested (getting responsees from finetuneed GPT)

The actual model wiegths could not be pushed to github

The dataset as well we have hidden for ethical reasons