OpenAI announces ‘Voice Engine’ to create human voices

/ Hotstar / By
Hotstar in UAE
Hotstar in UAE

OpenAI can now recreate human voices with its “Voice Engine” platform. The Microsoft-funded company is holding the new AI engine close to its chest owing to obvious ethical and legal implications.

After the massive success of ChatGPT, DALL-E, and Sora, OpenAI has now ventured into the field of sound, particularly, human speech and voice. The Voice Engine AI is currently in the beta testing phase but internet users might not be able to sign up for a subscription.

Voice Engine can recreate multilingual human voices with just 15-second recordings

ChatGPT took the world by storm, DALL-E unlocked creativity in millions, and Sora may soon allow ordinary internet users to become movie creators. After text, images, and videos, OpenAI has now announced Voice Engine, an AI platform that can recreate human voices.

The New York Times received a demo of the Voice Engine platform. The AI engine can reportedly recreate a human’s voice from a 15-second recording. If that’s not scary enough, after absorbing the needed information, Voice Engine can speak any text prompt into the recreated human voice.

Surprisingly, the text prompt need not even be in the native language of the speaker, whose voice was used to train the AI engine. In other words, a native English speaker could speak in Spanish, French, Chinese, or many other languages through Voice Engine.

Given the obvious pitfalls of such a technology, OpenAI has assured it is exploring multiple safety checks such as watermarks. There would be controls that restrict Voice Engine from creating the human voices of certain individuals.

Interestingly, OpenAI product manager, Jeff Harris has reportedly claimed the company doesn’t have any immediate plans to make money from the technology. The primary purpose of the is to be useful to people who lost their voices through illness or accident, he added.

Will OpenAI offer its latest creation to internet users as a subscription?

OpenAI’s Sora, which allows users to create a 60-second motion picture with just text prompts, isn’t available to the general public. Similarly, the company hasn’t released any version of Voice Engine.

Voice Engine is currently available to “a small group of businesses.” Simply put, a few companies, presumably by invitation, have limited access to the platform.

Needless to mention, there are huge ethical and legal implications of an AI platform that can recreate human voices, in multiple languages, based on a 15-recording.

OpenAI may have restricted access to Voice Engine. However, it is concerning to note that AI has metamorphosized into a behemoth.

In addition to convincing deep fake images and videos, scamsters and fraudsters have been using AI to mimic human voices for quite some time. Several victims have fallen prey to deep fake voice messages that sounded like their relatives were in trouble and were asking for money.

In the wrong hands, Voice Engine could accept and recreate the voices of politicians, celebrities, journalists, and other prominent personalities. The platform could be used to create convincing audio clips that spread misinformation or propaganda. In an equally scary scenario, hackers and criminals could compromise security systems that rely on voice authentication.

Hence, OpenAI would most likely heavily censor and regulate the Voice Engine platform. It would have to formulate some tough and smart digital locks as well as stringent safety checks to prevent its misuse.

Safety and ethical concerns aside, OpenAI’s Voice Engine could be tremendously helpful to film and web-series producers who need to dub their creations into other languages. Similarly, the educational and entertainment fields too could benefit immensely.

2024-03-30 15:04:47