Tech 6 months ago
Explore ChatGPT's lifelike advanced voice mode, offering real-time responses and engaging conversations. Discover how this upgrade redefines AI assistants.

OpenAI surprised users earlier this year by showcasing an updated voice mode for the advanced version of ChatGPT. Unlike the typical robotic voices associated with digital assistants like Alexa or Siri, this new ChatGPT advanced voice mode is surprisingly lifelike. It responds in real-time, can handle interruptions, makes giggling sounds when users joke, and gauges emotional tones. During the initial demo, it even sounded somewhat like Scarlett Johansson.

Starting Tuesday, the advanced voice mode will be available to paid users of ChatGPT-4o. It will first be rolled out to a small group of subscribers to the app’s “Plus” mode, with plans to make it accessible to all Plus users by fall.

While ChatGPT already had a less sophisticated voice mode, this new rollout marks a significant milestone for OpenAI. The advanced voice mode transforms ChatGPT into a more personal assistant, capable of engaging in natural, spoken conversations similar to talking with a friend. This ease of interaction could lead users to engage with the tool more frequently, challenging virtual assistant leaders like Apple and Amazon.

However, the introduction of a more advanced voice mode raises questions: Can the tool accurately understand users with speech differences? Will users place undue trust in a human-sounding AI assistant, even when it occasionally errs?

OpenAI had initially planned to roll out the advanced voice mode in June but delayed it by a month to ensure the tool’s safety and real-time responsiveness for millions of users. The company tested the AI model's voice capabilities with over 100 testers speaking 45 different languages from 29 different regions.

As a safety measure, the voice mode will be limited to four pre-set voices created in collaboration with voice actors to prevent impersonation. It will also block requests for generating music or copyrighted audio. Additionally, it will have the same protections as ChatGPT’s text mode to avoid generating illegal or harmful content.

One notable change from the May demo is that users will no longer have access to the voice that many thought sounded like Scarlett Johansson. Although OpenAI clarified that the voice was created with another actor's help and was not intended to mimic Johansson, they paused its use "out of respect" after the actor expressed concerns.

This launch comes on the heels of OpenAI's recent announcement that it is testing a search engine powered by its AI technology, further expanding its range of consumer-facing AI tools. The OpenAI search engine could potentially challenge Google’s dominance in the online search market.