Every new video showing off ChatGPT Voice's features makes me want to try it out for myself, and the latest one is no exception. In this video, we see the AI adopt different character voices based on simple voice prompts – perfect for storytelling.
It's not clear when the next version of ChatGPT Voice (aka Omni Voice) will be available, but rumors suggest the first users will have access to it later in the summer.
Unlike the current version of ChatGPT Voice, this new model is built using GPT-4o and natively synthesizes speech, without having to convert user speech to text first.
This native speech modality enables the model to create different sounding voices, express emotions, and even detect emotional signals in the voice as the user speaks.
What does the new ChatGPT demo show?
OpenAI has been slowly revealing a ton of hidden features in GPT-4o's new voice mode, which has so far been seen translating conversations in real time, helping with homework, and even greeting the audience at a French tech conference.
The latest demo begins with an OpenAI staffer giving instructions to an AI chatbot: He tells the AI he's writing a story and wants to practice voices for a few different characters, one of which is a lion, which ChatGPT provides with a gruff, commanding voice.
ChatGPT does a great job playing a lion, then can quickly jump to his second character, “a mouse who snuck into a cave.”
What was really interesting was how he instructed the AI to change its voice, “to squeak more like a little mouse.”
After that, we added other characters, such as an owl who acts as an advisor to the lion with a wise voice, and a villain character who has an evil laugh. ChatGPT has a maniacal laugh. This gave us a more complete set of characters to use in the story.
Overall it was a great result and gave us some insight into how ChatGPT could be used to leverage your Dungeon Master skills in a D&D game or to replace audiobooks with custom interactive stories generated on the fly.
When will ChatGPT Voice be available?
OpenAI clarifies that while voice mode is already available to all users of the ChatGPT app, “GPT-4o's new voice and vision capabilities will be rolled out in the coming weeks.”
Some users have started calling the new mode Omni Voice or GPT-4o Voice. The features showcased in the new video are only available in GPT-4o Voice and Vision. Some users will get access to it in the coming months.
When you go to our iPhone or Android app and enter voice mode, you can check which version you're using by clicking the (i) icon in the top right. If you're using the current version, you'll see the new ChatGPT Voice “Coming Soon.”
More articles from Tom's Guide
Back to mobile
