Tangentially related: as a non-developer I have started trying to hack/extend Anytime Podcast player to become a light language learning app, for supporting using native level resources for intermediate learners.

The idea is to give me vocab, phrases and idioms with timed with a transcript to appear just before the audio, with some feedback on understanding and a simple learning model of users level, vocabulary, strengths and weaknesses etc

I've made a start using whisper and some simple Bayesian modelling but if anyone more technically gifted wants to steal my idea they're more than welcome... As it's going to take a while for little old me.