Thanks for the tip! I will add GPT to the mix to clean up the speech and title data.