Continuing the discussion from Failed to download subtitle:
Seeing that the issue of downloading subtitles has been a recurring problem throughout Plex’s long history, I propose the consideration of using the neural net Whisper for on-the-fly subtitling and captioning. This can be accomplished via API. If that approach is not feasible due to overhead costs, then finding, funding, and supporting a free and open-source model that is compact and can run locally would be ideal—think of a transformer language model or another natural language processing (NLP) model.
Let’s not ‘stream’ around the bush; we need better subtitles, and we need them now! 
