OpenAI knows that AI-generated voice tools can be a fuzzy affair.
In a blog post sharing early results from the testing phase of its new synthetic voice tool, the artificial intelligence company raised concerns about using AI to reproduce human voices, particularly during an election year.
OpenAI’s “Voice Engine” tool, which the company says it first developed in late 2022, uses a 15-second audio clip of a real person’s voice to create an eerily realistic and human-like replica of that voice .
And users can make this voice say anything, even in other languages.
The tool is not yet publicly available and OpenAI says it is still considering “whether and how to deploy this technology at scale.”
“We recognize that generating speech that resembles the voice of the people carries serious risks, which are particularly significant during election times,” OpenAI wrote in its blog. “We are collaborating with U.S. and international partners in government, media, entertainment, education, civil society and beyond to ensure we incorporate their feedback as we build.”
OpenAI currently uses the tool to power ChatGPT’s “read aloud” features, as well as the company’s text-to-speech API.
Late last year, OpenAI began expanding the tool externally, working with what it described as a “small group of trusted partners” to test Voice Engine for things like hardware children’s educational, language translation and medical voice recovery, the company said. said in his message.
OpenAI stressed that its partner organizations must obey strict policies to use Voice Engine, such as obtaining consent from each spoofed individual and informing listeners that the voice is generated by AI.
“We are taking a cautious and informed approach to wider release due to the potential for misuse of synthetic voice,” the company wrote. “We hope to begin a dialogue about the responsible deployment of synthetic voices and how society can adapt to these new capabilities.”
Although the company said it does not yet know if it will ever release the tool to the general public, it has pushed policymakers and developers to take steps to prevent dangerous misuse of the technology that she was developing.
For example, OpenAI suggested establishing a “banned voices list” to prevent non-consensual replication of important voices, such as those of politicians or celebrities.
The company also recommended that banks stop using voice-based security authentication and that researchers develop techniques to determine whether a voice is real or fake.
businessinsider
