Artificial Intelligence: OpenAI offers audio cloning technology

The American company OpenAI has presented a new program for reproducing real voices. OpenAI displays On his website Many audio recordings created using Voice Engine. So, after uploading a 15-second audio clip of a human voice, the Voice Engine was able to imitate it, read text, and translate it as well. This would allow users, among other things, to create an audio recording of their voices in Chinese or French.

In September, the company, which also developed the chatbot ChatGPT, introduced HeyGen, a tool that can translate videos using artificial intelligence and adjust lip movements. According to OpenAI, HeyGen takes advantage of this technology From the audio engine.

Due to the risk of misuse, Voice Engine will not be accessible for free at first, according to the company. OpenAI wants to test the product with testers.

'Serious risks' in election year

“We recognize that generating rhetoric that sounds like people's voices carries serious risks that are particularly emphasized in an election year,” the company said. A new President of the United States is scheduled to be elected on November 5. Important elections are also taking place in several other countries this year.

In the US state of New Hampshire, authorities are currently investigating calls in which President Joe Biden's voice was imitated using artificial intelligence to deter people from voting in the local primaries in January.