In a surprising move, AI company OpenAI has unveiled new technology that can replicate a person’s voice based on just 15 seconds of recording. Despite showcasing this Voice Engine technology, the company has stated that it will not release it to the public due to safety concerns. This decision comes in light of potential misuse risks associated with generating speech that resembles real voices, especially during sensitive times like an election year.
OpenAI plans to offer a preview of the technology to early testers but has emphasized its commitment to not widely releasing it. These early testers have agreed to use the technology responsibly, ensuring that they do not impersonate individuals without their consent and disclose that the voices generated are AI-generated. This cautious approach aligns with OpenAI’s previous practices when unveiling new technologies, like the chatbot and image-generator DALL-E.
Some startups offer voice-cloning technology to the public or select business customers like entertainment studios, but OpenAI’s Voice Engine signifies a potential shift towards broader applications in speech recognition and digital voice assistants. This move could position OpenAI to compete with established voice products such as Amazon’s Alexa in the future.
OpenAI’s partnership with The Associated Press, facilitated through a licensing and technology agreement, enables the company to access part of AP’s text archives. This collaboration highlights the innovative potential of combining AI technologies with trusted news sources for diverse applications.
