June 9, 2025 – OpenAI has rolled out a major enhancement to its premium voice mode for ChatGPT subscribers, bringing notable improvements in vocal intonation, naturalness, and emotional depth to deliver a smoother and more “human-like” interaction experience.
According to OpenAI, the upgraded advanced voice mode has further refined the naturalness of speech, incorporating more nuanced pitch variations, realistic pacing—including pauses and emphasis—and precise emotional delivery across a range of feelings, such as empathy and sarcasm.
Additionally, the advanced voice mode now includes an intuitive and efficient multilingual translation feature. Users can simply request the voice to perform translation, and it will continue providing translation services throughout the conversation until the user stops or switches languages.
This update builds on the improvements made to ChatGPT’s voice mode earlier this year, which aimed to reduce speech interruptions and refine accents.
During testing, OpenAI observed that the update might occasionally lead to slight degradation in audio quality, including unexpected changes in tone and pitch. These issues were more pronounced in certain voice options, but the company expects audio consistency to improve over time. Moreover, despite the upgrades, a small number of “hallucinations” remain in the voice mode, which may produce unexpected sounds resembling advertisements, gibberish, or background music. The development team is actively investigating these issues and is committed to finding solutions as quickly as possible.
