Post #4024507

2025-08-22 09:21 UTC

You see, unlike other Apple devices - Apple does not allow you to use #Siri as a system voice in VisionOS so the #TTS output from the agent is #robotic and doesn't sound natural at all... So I decided to include on device TTS using Chatterbox's TTS model for MPS (Apple Silicon) to make awesome sounding voice responses. That still wasn't enough...

Replies (1)

@thatprivacyguy@eupolicy.social 2025-08-22 09:24

I got to thinking - you know, it is really annoying when Siri mistakes something I say as a wake phrase - so I decided to take a different approach. I was thinking, why do we go through the trouble of parsing audio input using NLP to find out if it contains a wake phrase - that seems like a lot of steps that could be improved... So now I am working on creating a wake input which can be based on any audio signal (words, a song, a whistle a click - any sound).

Open ##4024508