One thing these trained voices make clear is that it's a tts engine generating ChatGPT-4o's speech, same as before. The whole omni-modal spin suggesting that the model is natively consuming and generating speech appears to be bunk.
I'm not familiar with the specifics of how AI models work but doesn't the ability from some of the demos rule out what you've said above? Eg. The speeding up and slowing down speech and the sarcasm don't seem possible if TTS was a separate component