OpenAI Unveils Advanced Voice Mode: Enhanced Voices and a Fresh New Design

OpenAI has announced the launch of Advanced Voice Mode (AVM) for a broader range of ChatGPT’s paying subscribers as of Tuesday. This innovative audio feature enhances the conversational experience, making interactions with ChatGPT feel more natural. Initially, it will be available to users within the ChatGPT Plus and Teams tiers, with Enterprise and Edu customers set to gain access next week.

As part of this update, AVM has undergone a design refresh. Users will now see a blue animated sphere representing the feature, replacing the previous black dot animation showcased in May.

Users will receive a notification in the ChatGPT app, positioned next to the voice icon, when AVM becomes accessible to them. The rollout will occur throughout the week for all Plus and Team subscribers.

While you've been waiting for this exciting update, we've also added Custom Instructions, Memory capabilities, five new voices, and enhanced accent recognition. Notably, ChatGPT can now say "Sorry I’m late" in over 50 languages.

In addition to these enhancements, ChatGPT introduces five new voices: Arbor, Maple, Sol, Spruce, and Vale. This addition brings the total number of voices to nine, closely approaching the offerings of Google’s Gemini Live. You might notice that all these names draw inspiration from nature, reinforcing AVM's goal of creating a more organic interaction with ChatGPT.

However, one voice absent from this selection is Sky, which OpenAI had highlighted during its spring update. After a legal challenge from Scarlett Johansson, who claimed Sky's voice was reminiscent of hers, OpenAI swiftly retracted this option, asserting that the resemblance was unintentional.

Another feature not included in this rollout is ChatGPT's video and screen-sharing capability, introduced during the spring update. This functionality was designed for GPT-4o, allowing it to process visual and audio information simultaneously. In demos, users could ask ChatGPT real-time questions about math problems on paper or code displayed on a screen. Currently, OpenAI has not provided a timeline for the release of these advanced multimodal features.

While OpenAI has made improvements since the limited alpha testing of AVM, including better accent recognition and smoother conversation flow, some glitches were present during our tests. Nevertheless, the company claims these issues have been addressed.

Additionally, OpenAI is enhancing ChatGPT’s customization features for AVM users, which include Custom Instructions for personalized responses and Memory, allowing the AI to retain and reference past conversations.

An OpenAI representative mentioned that AVM is not yet available in several regions, including the EU, the U.K., Switzerland, Iceland, Norway, and Liechtenstein.