10 Comments
Comments from the original post on X (Twitter)

@isidentical
looks 🔥
4 likes

@mayfer
how is this different from whisper.cpp's metal/coreml implementation?
3 likes

@m1guelpf
oooo this is really cool!
2 likes

@AlexBerriche
Super cool !
1 like

@jimmygunawanapp
I tried it so far I like the realtime streaming and translation. I test it while YouTube running on TV and translation works 50-60% some times it's buffering. Can it tend later audio from say running audio app like YouTube or Twitter video?
1 like

@TomMaixner
I have problems without loading
1 like

@apanasenko
Any plan for diarization?
1 like

@argmaxinc
Yes
3 likes

@priontific
Hi! 👋 Quick qu; is my understanding right that WhisperKit processes live audio in chunks and stitches the transcript together? Or does it keep a run-on KV cache of encoded audio which grows as more audio is streamed in + encoded?
1 like

@beats193
@osanseviero This is so cool! Can you please add an action in iOS shortcuts app so we can integrate it with other iOS native apps?