I tested ChatGPT's Live Voice upgrade, and it almost felt human - how to try it
TL;DR
ZDNET tested ChatGPT’s new Live Voice upgrade: the latest GPT-Live models can listen, speak, and do online research while the conversation continues. The experience feels smoother because ChatGPT waits through pauses more naturally and interrupts less than older voice mode versions. The human feeling is still an almost: voice, timing, and follow-ups are stronger, but the assistant remains recognizably synthetic and polished.
Nauti's Take
This sounds like a real upgrade, not magic. Fewer interruptions and parallel research fix two annoying voice problems, but they do not make ChatGPT human.
The important question is not whether it feels like a person, but whether users can still tell in a live conversation when the system knows something, is guessing, or is simply smoothing over uncertainty.
Briefingshow
Voice is moving from novelty to interface. If an assistant can research, translate, and keep listening while talking, AI use shifts away from typed prompts toward continuous conversation. That makes trust, control, and clear boundaries more important than the polish of a demo.
Sources