April 2, 2024: OpenAI, the company behind the popular ChatGPT, has announced Voice Engine, a new text-to-speech AI model that can generate synthetic voices based on just 15 seconds of recorded audio.
Developed in late 2022, the technology has the potential to offer a number of benefits, including assisted reading, expanded global reach for creators, and personalized voice options for those who cannot speak.
But despite these potential benefits, OpenAI has decided not to release the technology publicly for now due to concerns about potential misuse.
The company originally planned to launch a pilot program earlier this month that would allow developers to sign up for its voice engine API, but scaled back its ambitious plans after considering the ethical implications.
“We are releasing this technology in preview now, but will not be releasing it broadly,” OpenAI said in a statement. “We hope this preview of the Voice Engine highlights its potential and serves as a reminder of the need to build societal resilience to the challenges posed by more compelling generative models.”
The company has been testing the technology with select partners since last year, requiring them to agree to terms of use that prohibit unauthorized impersonation and to give informed consent for the individuals whose voices are being copied.
OpenAI also implemented watermarks on all speech samples to help track the origin of every speech generated by their speech engine model.
To address the potential risks associated with voice-cloning technology, OpenAI made three recommendations for society to adapt: phasing out voice-based authentication for bank accounts, educating the public about the potential for fraudulent AI content, and accelerating the development of technologies to trace the origin of audio content.
The company emphasizes the need for a cautious and informed approach to the widespread rollout of synthetic voice technology.
“We want to start a conversation about the responsible deployment of synthetic voices and how society can adapt to these new capabilities,” OpenAI said. “Based on these conversations and the results of these small-scale tests, we will make more informed decisions about whether and how to deploy this technology at scale.”
As development of voice-reproduction technology continues to advance, it is important for companies like OpenAI to consider the potential risks and ethical implications while working to harness the benefits of this technology for society.