OpenAI announces a risky voice cloning AI model 'Voice Engine'

OpenAI has announced a new AI voice cloning model called Voice Engine. However, the public release is not happening any time soon due to associated risks. 

OpenAI’s new voice cloning AI model can recreate a person’s voice with a 15-second audio sample

On Friday, OpenAI announced its new custom voice creation model – Voice Engine. It is worth noting that OpenAI filed a trademark application for the model name a few weeks back.

Talking of the same in its blog post, OpenAI says the new AI voice cloning technology can recreate a natural-sounding person’s voice with just 15 seconds of audio recording of the person.

OpenAI opted to preview the newly announced Voice Engine and not widely release it, at least for now. It also mentioned that this move is an attempt to curb the risk of misinformation that could affect the US elections. A part of the statement from the company reads:

We recognize that generating speech that resembles people’s voices has serious risks, which are especially top of mind in an election year. 

The AI giant said the early Voice Engine testers agreed to their usage policies. It prohibits early testers from impersonating individuals or entities without their consent or legal rights. Moreover, they need to disclose that the voices are AI-generated. 

More recently, citing similar risks, the OpenAI CTO Mira Murati confirmed the company’s plan to push the SoraAI public release after the presidential elections. It seems OpenAI is well aware of the risks AI models bring into the picture. 

Will Voice Engine be the next big thing from OpenAI? Let us know your thoughts in the comments below.