OpenAI's Voice-Cloning Tool Needs Just 15 Seconds of Audio

OpenAI has developed a tool it admits is too risky to be released to the general public right now. The ChatGPT maker launched a "small-scale preview" of Voice Engine, which it discusses in a blog post here. Some takeaways:

• The bottom line: "We are taking a cautious and informed approach to a broader release due to the potential for synthetic voice misuse," the blog post explains. "We hope to start a dialogue on the responsible deployment of synthetic voices, and how society can adapt to these new capabilities." Those conversations, plus the results of more "small scale tests," will ultimately lead to "a more informed decision about whether and how to deploy this technology at scale."

• What it can do: Voice Engine requires just one 15-second audio sample which, along with text input, is used to "generate natural-sounding speech that closely resembles the original speaker" and uses "emotive and realistic voices," the blog post says.
• A look at the comments section: In the replies to a post on X regarding Voice Engine, responses like this could be found: "What could possibly go wrong?"
• Already in use: As the Verge explains, Voice Engine has already been used to power ChatGPT's Read Aloud feature, among other things. A member of the product team previously said it was trained using "a mix of licensed and publicly available data."
• Some potential other uses: The Guardian points out a few ways partners used Voice Engine in the small-scale test, including scripted voiceovers for education technology and translations that preserve the accent and voice of the speaker.

(More OpenAI stories.)