I also noticed the voice, but it sounded really weird when the robot started answering with something that sounded like hesitation. Is it really THAT real? I cannot imagine a computer hesitating before starting a sentence.
Hesitation is a personafication. Kruel.ai is built the same and if you use chatgpt with voice the same there. Using pauses , even neuances like ah and um. These are by design to make them feel more human 
Thanks for the clarification. Really interesting where we are getting to…
Stuttering perfected it…
Very nice …
This is the cutest thing I ever saw!