Every new video showing off the features of ChatGPT Voice makes me excited to try it out for myself, and the latest one is no exception. In it, we see ai adopting different character voices based on simple voice prompts.

It's not clear when the next version of chatgpt Voice, also known as Omni Voice, will be available, but rumors suggest that the first users will have access later in the summer.

Unlike the current version of ChatGPT Voice, this new model is built using GPT-4o and speaks first because it converts voice to speech natively.

This native voice modality allows the model to create a different sounding voice, express emotions and listen to your voice while you talk to it. It is what makes it possible to detect signs of emotion in the voice.

OpenAI has gradually revealed a number of features hidden within the new voice mode of GPT-4o. So far, I've seen them translate conversations in real time, help with homework, and greet their audience at a technical conference in France.

The latest demo opens with OpenAI staff giving instructions to the AI chatbot. He says he's writing stories to AI and wants to practice some voices for different characters. One is a lion, and ChatGPT puts on a sullen, majestic voice.

ChatGPT does a great job of the lion, allowing him to quickly jump to the 2nd character, the "mouse that crept into the cave."

What was really interesting was how to tell the AI to change their voice and "make it squeak a bit like a little little mouse."

Then he added another character, a villain character that sounds smart like an owl, acts as an advisor to a lion, and is accompanied by an evil laugh. ChatGPT gave a maniac laugh! It created a rounded set of characters for use in the story.

Overall it did a great job and gave us a chance to see how ChatGPT could potentially act as Dungeon master in D&D games or custom generated on the fly

OpenAI said that all users of the chatgpt app already have voice mode available, but "the new voice and visual mode of GPT-4o is not available." It is clear that the feature will be deployed in the coming weeks

Some users are starting to call the new mode Omni-Voice or GPT-4o voice. The features shown in the new video are only available in GPT-4o voice and vision. Some users will get access in the coming months.

When you enter the iPhone or Android app and enter voice mode, you can click the (i) icon in the upper right corner to see which version you are using. It should say "Coming soon" if you're in the current version, the new ChatGPT voice.