ChatGPT has gone full virtual assistant with voice and image recognition

Get ready to hear what AI is thinking. ChatGPT is gaining voice and image recognition capabilities on top of a voice of its own, but you’ll have to pay to hear it.

As we know chatbots, conversation is generally spurred through text, not audible inquiries. Services like Bard, ChatGPT, and Bing Chat all follow these models, and we were just starting to consider all of this normal.

ChatGPT is uprooting that with a new update that brings an audible voice to the advanced language model. With that, it can read aloud its responses in a wildly natural voice that rivals every other virtual assistant currently available. On your end, you can use your human voice to ask questions and make conversation like you did before with text.

In essence, ChatGPT now has a natural-sounding virtual assistant that can read your bedtime stories and respond with all the unhinged things it did before, but in a soothing tone. OpenAI posted a video of a demo interaction, and it’s wild to see in action. ChatGPT’s voice is natural and calm, creeping closer to that boundary where virtual speech becomes indistinguishable from that emitted of vocal cords.

OpenAI notes that this is accomplished via voice actors, though it still takes a lot of effort to produce these synthetic voices out of sample audio.

In addition to voice, you’ll be able to take and upload photos and ask questions, sort of how Google Lens operates. Not only does it recognize the subject of the photo – you can circle items to focus its attention – but it will also let you take additional photos to provide context.

Again, a ChatGPT demo video goes along with this new image recognition feature. In the video, the human asks about a bike seat and how to adjust it. After a couple of photos, ChatGPT knows what needs to be done and says as much. The most impressive part of this all is that ChatGPT was able to read the screenshot of the bike’s manual and direct the person to the correct tool – 4mm Hex key – in his basic toolbox.

To get access to these new features, you’ll need to be a ChatGPT Plus or Enterprise member. Plus, it starts at $20/month and will set you in line to get this upcoming update, which should roll out in the next couple of weeks. OpenAI noted that “other groups of users” will see the features soon after, though it’s unclear if that pertains to free tiers.