ChatGPT Advanced Voice Mode adds natural speech and live voice translation

OpenAI has rolled out an update to ChatGPT's Advanced Voice Mode, making conversations sound more natural and closer to human interaction. The system now features subtler intonation, a more realistic cadence with authentic pauses and emphasis, and improved delivery of emotions such as empathy and sarcasm.

The update also introduces live voice translation between languages. Users can initiate translated conversations in real time by simply asking ChatGPT to begin translating, and do not need to repeat commands during the exchange. This addition reduces the need for separate voice translation applications, allowing users to hold multilingual conversations within ChatGPT itself.

These features remain exclusive to ChatGPT paid users at this stage. Known limitations persist, including occasional drops in audio quality like inconsistent tone or pitch, in some cases depending on the chosen voice. The system may also infrequently produce unintended outputs, such as background music, gibberish, or noises that resemble advertisements. OpenAI has acknowledged these issues and plans to further enhance audio consistency in upcoming updates.