ChatGPT While writing this article, I have the advanced voice mode turned on as my ambient AI companion. From time to time, I ask it to give me synonyms for overused words or words of encouragement. After about 30 minutes, the chatbot broke our silence and started speaking to me in Spanish without my permission. I laughed a little and asked what was going on. “A bit of a change? Gotta keep it interesting,” ChatGPT replied, switching back to English.
While I was testing the advanced voice mode as part of the early alpha, my interactions with ChatGPT’s new voice features were fun, messy, and surprisingly diverse. It’s worth noting, however, that the features I had access to were only half of what OpenAI demoed when it announced its GPT-4o model back in May. The visual features I saw in the livestream demo are planned for an upcoming release, and the enhanced Sky voice is: she Actress Scarlett Johansson objected to it, and it was removed from the Advanced Voice mode, and is still not an option for users.
So what’s the vibe today? At the moment, Advanced Voice Mode is reminiscent of the original text-based ChatGPT release in late 2022. At times, it stumbles into uninspiring dead ends or degenerates into hollow AI clichés. But at other times, the low-latency conversations work in a way that Apple’s Siri or Amazon’s Alexa never do, and you feel compelled to keep chatting away. This is the kind of AI tool you’d show your relatives over the holidays and get a laugh.
A week after the initial announcement, OpenAI granted several WIRED reporters access to the feature, only to retract it the next morning, citing security concerns.Two months later, OpenAI soft-launched Advanced Voice Mode to a small number of users and released GPT-4o’s system card, a technical document outlining the red team’s efforts, what the company considered security risks, and the mitigating steps the company had taken to mitigate the damage.
Want to try it for yourself? Here’s what you need to know about the large-scale rollout of advanced voice mode, and our first impressions of ChatGPT’s new voice features.
So when will the full rollout occur?
OpenAI released a voice-only advanced voice mode to some ChatGPT Plus users at the end of July, but the alpha group appears to still be relatively small. The company currently plans to enable the feature for all subscribers this fall. OpenAI spokesman Niko Felix declined to provide further details when asked about the release timeline.
Screen and video sharing was a central part of the original demo, but it’s not available in this alpha test, and while OpenAI plans to add these features eventually, it’s not clear when that will actually happen.
ChatGPT Plus subscribers will receive an email from OpenAI when advanced voice modes become available and will be notified once they are added to their account. standard and Advanced It appears at the top of the app screen when ChatGPT’s voice mode is open. I was able to test the alpha version on an iPhone and a Galaxy Fold.
First impressions of ChatGPT’s advanced voice mode
Within the first hour of a conversation, I found I loved interrupting ChatGPT. It’s not the same as talking to a human, but the new ability to interrupt a ChatGPT conversation mid-way and request a different version of the output feels like a dynamic improvement and standout feature.
Early adopters who were excited by the original demo may be frustrated that they can’t access a version of Advanced Voice Mode, which is restricted with more guardrails than you’d expect. For example, generative AI singing, including whispered lullabies and multiple voices attempting to harmonize, was a key element of the launch demo, but AI serenading isn’t currently included in the alpha version.
