OpenAI develops Voice Engine, a controversial AI tool capable of cloning a voice from a 15-second sample, sparking discussions on responsible use and security risks. In a blog post, OpenAI revealed that it recently conducted a small test of a model called Voice Engine.
Voice Engine is a new technology that OpenAI has been developing. The tool's functionality is highly contentious and, if disclosed publicly, might provide a number of security issues. The Voice Engine, which is an AI tool, can listen to a voice sample for only 15 seconds, and then clone it. Meaning, it is capable of producing the same voice just by listening to it for such a short duration. This is one of the primary reasons why OpenAI has not made the feature public yet.
OpenAI disclosed in a blog post that it has carried out a brief trial of a Voice Engine model. With text input and a 15-second audio clip of a human voice, this model can produce speech that closely resembles the actual speaker. It's amazing that this algorithm can produce voices that sound authentic and emotive from such a little sample.
Originally created in late 2022, Voice Engine is currently powering preset voices in text-to-speech programs like as Read Aloud and ChatGPT Voice. OpenAI is cautious about making it broadly available, though, as there's a chance that individuals may abuse artificial voices. They want to have a discussion about how society can accept these new skills and how to use synthetic voices ethically.
"We acknowledge that producing speech that sounds like people's voices has significant hazards, which are particularly important during an election year. To make sure we are incorporating their feedback as we build, we are engaging with U.S. and international partners from across government, media, entertainment, education, civil society, and beyond," the blog added.
Despite its benefits, there are concerns about misuse, especially during election seasons. OpenAI is working closely with various partners and implementing safety measures to prevent misuse.