OpenAI’s Voice Engine Can Clone Your Voice Within 15 Seconds

OpenAI has just unveiled news of a limited-scale preview for its latest tool, Voice Engine. This technology, referred to as voice cloning, is capable of replicating any speaker’s voice by analyzing a mere 15-second audio snippet. The company asserts that it produces speech that sounds natural, complete with emotive and realistic tones.

The OpenAI Voice Engine has been in the works since 2022 and is built upon the company’s existing text-to-speech API. OpenAI has already integrated a variant of this toolset to drive the preset voices featured in the existing text-to-speech API and the Read Aloud function. A series of samples showcased on the company’s official blog reveals remarkably authentic renditions, bearing striking resemblance to genuine human voices.

This technology should help provide reading assistance and language translation, and help those with disabilities, according to OpenAI. The technology is already helping a patient with speech impairment using a Voice Engine clone pulled from audio recorded in a school project. This was done through a Brown University pilot program.

But with all its potential benefits, there is also a cause for concern when it comes to bad actors. Voice Engine could easily be used to create highly realistic deepfakes of someone, a trend that is already prevalent in the industry. This shows that Voice Engine is not fully ready for a public release yet and some privacy issues have to be addressed still.

OpenAI acknowledges this issue, saying that the tech has “serious risks, which are especially top of mind in an election year.” The startup is currently in the process of taking in user feedback from “US and international partners from across government, media, entertainment, education, civil society and beyond.”

OpenAI is yet to reveal a release date for Voice Engine.

Get Alerts

Follow ProPakistani to get latest news and updates.

ProPakistani Community

Join the groups below to get latest news and updates.