Advanced Voice Mode is an innovative feature for ChatGPT that allows users to engage in real-time, conversational interactions with the AI chatbot, eliminating the need for text prompts or lengthy audio exchanges. Launched in late July for selected Plus subscribers, this feature was first showcased at OpenAI’s Spring Update event.
OpenAI describes Advanced Voice Mode as offering “more natural, real-time conversations” and enabling users to interrupt seamlessly. Additionally, it detects and responds to users' emotions and can even take breath breaks and mimic human laughter during chats. If you haven't gained access yet, don’t worry—it's rolling out to more users soon.
Recently, OpenAI officially introduced its long-awaited Advanced Voice feature to a limited group of ChatGPT Plus subscribers, making it available for some to explore. While the exact size of this initial rollout remains unclear, the company has committed to expanding access in the coming weeks, with full availability expected for all Plus subscribers by this fall. Although many users are eager to try it, you can anticipate access anytime before winter, unless further delays occur. You’ll know you have access when you receive an email invitation or a notification in the ChatGPT app.
To utilize Advanced Voice Mode, users must have a Plus subscription and an Android device with app version 1.2024.206 or later, or an iPhone running iOS 16.4 or later alongside the same app version. It’s important to note that having the right device does not guarantee participation in the alpha release phase. OpenAI has not disclosed the criteria for selecting users for this feature, but selected individuals will receive both an email notification and a tooltip in the ChatGPT mobile app to access the new mode.
During the alpha phase, OpenAI will collect audio from conversations using Advanced Voice Mode to enhance its models, provided that users have not opted out of data sharing. To disable this option, navigate to the Data Controls tab in your app's settings and uncheck "Improve voice for everyone."
OpenAI has stated that both input and output for Advanced Voice have daily usage limits, although specific durations have not been disclosed, and those limits may change over time. Nonetheless, users such as Himels Tech have demonstrated conversations lasting nearly 10 minutes. The AI will notify users when they have three minutes remaining, concluding the chat and reverting to the standard voice interface.
At its core, Advanced Voice Mode provides a new method of interacting with the existing GPT-4o large language model, enabling users to utilize it for various tasks. In essence, anything achievable with text-based ChatGPT is possible with Advanced Voice, enhanced by its amusing vocal features. Early adopters are exploring its capabilities, from beatboxing to storytelling and rapid counting.
However, there are safety measures and limitations in place for Advanced Voice Mode. Users cannot create memories, utilize custom instructions, or access GPTs in this mode. While it can remember details from previous Advanced Voice conversations, it cannot reference earlier chats conducted through text prompts or the standard voice interface.
Additionally, Advanced Voice will not perform singing, regardless of requests. According to OpenAI, “to respect creators’ rights, we’ve implemented several measures, including new filters, to prevent Advanced Voice Mode from producing musical content such as singing.”