lesssgooooo!

Cool project and impressive comparative benchmark results, but at least let's put some info about what it is in this post:

Open, affordable, and production-ready speech understanding for everyone.

It's speech recognition/comprehension AI, as far as I understand.

Sweet, perhaps it will run better than Whisper (according to the graphs at least) on my poor phone as voice input method. Whisper works great if I give it 20-30s to think :)

And it still fucks up one word in twenty. But... one word in fifty, for Italian? Damn, guess colleges should have stuck with Latin.

Neopolitans still can't use it until they implement gesture recognition.