Earlier quoted context omitted.
One of the millions of features I had penciled in for my podcasting app - that didn't survive a rewrite in typescript - was the ability to detect different speakers and skip to the next one. Probably even topic based could work too if you can convert the audio to words at specific instances, subject detection may or may not work well I guess. Worth a try.
I actually know someone developing technology for this use case. She's hearing impaired and is trying to build an application that transcribes voice to text from a variety of people in a single conversation. Very helpful for her in social situations where she can't lipread multiple people at once, especially if they're wearing masks obviously. https://imanyco.com/
Re: Show HN: My new platform. Check it out
#11Cool stuff. The pioneering efforts of a11y tools can inform UI patterns in ways that go well beyond accommodating the specific needs they were invented for. Having witnessed a blind person using VoiceOver on her iPhone with a very high speed (like, the 2x or 4x setting on many video players doesn't even come close to the speed I'm talking about!) actually contributed to the idea I presented in my earlier comment: consuming audio in a "speed reading" fashion. Nobody's got the patience to drink from the social media firehose at conversational speed.