OpenAI created a voice cloning tool but delayed its launch for security reasons

Sam Altman’s OpenAI technology company has been responsible, in recent years, for implementing Artificial Intelligence in the daily lives of millions of people around the world. And there’s news.

The famous ChatGPT language model is now joined by Voice Enginea new type of technology that allows clone anyone’s voice, using Artificial Intelligence. Through a 15-second audio and some written text, it is possible to simulate the verbal interaction of a human being “with a natural sound that closely resembles the original speaker”, OpenAI advances. But the high risks associated with voice cloning led the company, which is based in San Francisco, to postpone its launch. The objective is to avoid a possible increase in misinformation, in a year when several countries in the world are going to elections, such as the United States, Russia, India or Mexico, with European elections still taking place.

O Voice Engine was developed in late 2022, OpenAI revealed, and a year later it was tested “privately with a small group of trusted partners.” However, the technology company chose to postpone its launch and preferred to adopt an “informed” approach. and cautiously” to ensure that the tool is not used improperly, can be read in the official announcement.

“We hope to start a dialogue about the responsible use of synthetic voices and how society can adapt to these new capabilities”, says Sam Altman’s company. After the conversations are concluded and the results are presented, a decision will be made “more informed about whether and how to implement this technology at scale”.

The most recent tests carried out on Voice Engine by the various partner companies in the project resulted in the discovery of a set of benefits associated with the use of the new technology. According to OpenAI, this tool can be used as a reading aid for children or people who do not know/can read. Also in translation videos, podcasts or other content, preserving the characteristics of the original speaker. It can also be used to facilitate communication among people with disabilities or learning needs, as well as allowing patients who have lost the ability to speak for health reasons.”recover your voice”.

“Partners who test the Voice Engine have agreed to our terms of use, which prohibit impersonation of another individual or organization without consent or legal right,” guarantees OpenAI. Likewise, users must be informed that the voices are generated by Artificial Intelligence and, for security reasons, all audio generated by the Voice Engine will have a watermark to facilitate identification and monitoring of the final use of the audio.

SAM ALTMAN’S COMPANY IS NOT THE FIRST TO CLONE VOICES

ElevenLabs – considered leader in this area – is a company specializing in voice cloning and synthesis in different languages ​​using Artificial Intelligence. The start-up that was founded in 2022 has as its main mission “to make content universally accessible in any language and voice”. In January the start-up increased its capital by around 80 million dollars and its main clients include companies linked to the advertising and audiovisual sector.

Recently, ElevenLabs announced a partnership with the star-up Rabbit, responsible for launching “R1”, one portable artificial intelligence assistant. This little gadget was first presented at this year’s edition of Consumer Electronic Show (CES) in Las Vegas and according to the company itself It has already sold 100,000 units since its launch.

Article by Mariana Jerónimo, edited by Mafalda Ganhão