The new AI model can clone voice from just 15 seconds of audio. Recently, artificial intelligence company OpenAI unveiled a new voice cloning AI model called Voice Engine.
According to OpenAI, this model can generate a voice similar to a real speaker from a 15-second audio sample in addition to text.
OpenAI has made the voice engine available to a select group of beta testers for now, keeping in mind potential abuse and associated risks.
“We recognize that there is a serious risk of creating rhetoric that resembles the voice of the public, which is especially top of mind in an election year,” the San Francisco-based company said in a statement. In sensitive contexts such as elections, there is growing debate over impersonation of individuals’ voices. The robocalls are allegedly producing AI-generated voices impersonating political figures.
Note that several startup companies are already offering voice-cloning.