ChatGPT’s Advanced Human-like Voices Are Finally Here

• ChatGPT finally released its set of 4 advanced voices this Tuesday.
• It will first be available only to a selected set of Plus users. And by this fall, it will be made available to all Plus users.
• Although the ChatGPT had an existing voice mode, the new and advanced voices are faster and more realistic.

ChatGPT's advanced voice modes are finally here. The company started rolling them out from Tuesday onwards. It will first be made available to a small group of users subscribed to the Plus" package.

These users will get an alert on their ChatGPT app followed by an email with instructions on how to use it. And then based on their feedback, it will be rolled out to all Plus users soon in the fall of 2024.

What Makes ChatGPT's Voices Different?

Although ChatGPT already has a voice mode, the advanced voices are just different. The older model was very slow because it broke down the process into 3 steps:

• In the first stage, the voice of the user is converted to a text that ChatGPT can understand.
• In the second stage, the prompt is processed.
• In the last stage, ChatGPT's text is converted into audio.

But ChatGPT-4o on the other hand, which hosts the new advanced models, can process all these tasks without the help of auxiliary models. Hence, it's much faster.

And unlike the robotic voice of Siri, Alexa, and Google Assistant, ChatGPT's voices are very life-like. They answer in real-time, can adjust to interruptions (just like a real human conversation), giggle at your jokes, and change their tone based on the user's emotions.

People who have watched the movie Her" starring Scarlett Johanson as an AI will have a better idea of what these voices sound like. In fact, one of the voices called Sky" that was showcased during the demo sounded exactly like Scarlett Johanson from the movie.

The Controversy on ChatGPT's AI Voices

The similarity of the two voices led to a lot of controversy. After some users pointed out the uncanny resemblance, Johanson said that OpenAI's CEO Sam Altman apparently reached out to her a number of times to use her voice. However, she declined every single time.

So after the demo, she hired a legal counsel to prevent her voice from being used in ChatGPT. Although Altman had said time and again that Sky" is not Johanson's voice, it was ultimately removed in June.

As of now, the four official ChatGPT voices are Juniper, Breeze, Cove, and Ember.

At the same time, he had also announced a slight delay in the launch of the advanced voice mode over safety issues. The company wanted to run more tests and tune some features to ensure the voices would act appropriately.

• OpenAI has revealed that it has prevented people from adding more voices to avoid impersonation.
• It also blocked the tool from processing certain requests such as those aimed to generate copyrighted music.

Reportedly, these voices were tested with more than 100 external red teamers who speak 45 different languages. A detailed safety report will also be published in August.

The post ChatGPT's Advanced Human-like Voices Are Finally Here appeared first on The Tech Report.