> (...) autogenerate captions (...) is fast but not always accurate and often misspells common words. (...) There’s also no way to stylize your captions

Nowadays, whisper models also run on device, put tend to handle this pretty well. I'm wondering why they had to go for their own (lesser) model.

Do anyone know if they auto-detect the language in the video and also generate captions for other languages? (just like whisper models do).

I don't use either FCP or Logic, but I am an avid user of Pixelmator who was pretty worried that the Apple acquisition meant it would up in a graveyard of abandoned Pro apps. This is some encouraging evidence that Apple does still care about these apps and that may not happen, although they really need to do a better job communicating this to people. Apple's trademark secrecy really isn't doing them any favors in spaces where people are constantly worried that the software they rely on is abandonware, and the utter lack of communication from Apple isn't doing anything to dispel these worries.

Honestly this is a really good update.