>>95902707Have you seen Whisper in action from OpenAI? It's really impressive and I have a setup to use it to watch Vtubers live with subtitles by extracting the audio, converting it into chunks, and feeding it into the AI. I pretty much never saw it misinterpret Mint's words during her playthrough except when the audio cut off mid-word and it got it right even then like 70% of the time.