You can now use ChatGPT’s advanced voice mode with real-time responses

ChatGPT with Advanced Voice Mode is now rolling out to some paying users, the company announced on Tuesday. OpenAI first announced the feature during its Spring event in May. Powered by the latest GPT-4o artificial intelligence (AI) model, OpenAI’s Advanced Voice Mode offers features such as real-time responses, natural voice, and the ability to sense the user’s emotions. The company said that all ChatGPT Plus users will get the feature in the fall of this year. However, there’s no word on when the video and screen sharing features, which were also demoed during the event, will be released.

OpenAI rolls out advanced speech mode for ChatGPT

OpenAI announced the rollout of ChatGPT’s advanced voice capabilities in a post on X (formerly known as Twitter). The company highlighted that the new voice mode allows users to interrupt the AI ​​chatbot at any time and offers a more natural interaction with voice modulations. It also shared a short video explaining how to enable the feature once it’s active.

According to the video, select ChatGPT Plus users will see an invitation notification at the bottom of the screen asking them to try the Advanced Voice Mode after opening the app. Tapping on it will take users to a new page titled “You’re Invited to Try the Advanced Voice Mode” and a button to activate the feature.

The feature is currently available to a small group of Plus users, but the company has not specified criteria for participation. The feature, which is described as an alpha rollout, is powered by OpenAI’s latest flagship large language model (LLM), GPT-4o.

The AI ​​firm explains the reason behind the delay said“Since we first demonstrated advanced voice mode, we’ve been working to improve the security and quality of voice conversations as we prepare to bring this groundbreaking technology to millions of people.”

OpenAI also highlighted that GPT-4o’s speech capabilities were tested with more than 100 external red teamers in 45 languages. Red teamers are cybersecurity professionals tasked with testing the security of a product or organization by simulating cyberattacks and jailbreak attempts. The goal of the process is to expose vulnerabilities in the system before it goes live.

At the moment, you’ll only have access to four preset voices after the feature rolls out to your account. Sky, the controversial voice that reportedly bore a strong resemblance to actor Scarlett Johannson, has yet to be added to ChatGPT.