Dunk said he uses "whatever the default streamelements tts voice is". That was not enough information,
so with some trial and error, I've figured out that it's Madison (English, American)
voice model
from the Amazon Polly TTS service.
You can test it on something like this https://lazypy.ro/tts/?voice=en-US-Wavenet-F&service=StreamElements&text=&lang=English&g=A
Compared to a legend like Brian, Madison is very hard to find workarounds/breaks/funny sounds for (or I just suck at this).