nice video on an enby's experience with being queer and neurodivergent through their gender expressions and problems with holding conversations and social anxiety. they use a text-to-speech system to talk as they prefer using it over their actual voice
I was thinking what if you did VOICE to TEXT to VOICE
That would be good for faster paced games where you cant type. You can say things literally however you want and then it comes out as the flat AI voice on the other end
I may try to do something like this in my free time because it actually sounds like something I'd use. There's public libraries & models (OpenAI's Whisper comes to mind) available that seem pretty mature which could be a good starting position for the speech recognition part.