There's a tool called SpeakerSplit here's their info: https://speakersplit.io/#/about it uses AI to separate the speakers on an audio track into two separate files. It also does transcription/diarization so you know which speaker is talking. I've tried them out on Notebook LM podcasts and it saved me an 1hr of editing, to like 2 minutes! The service is not free, but its very cheap.