OpenAI – creator of ChatGPT – revealed this Friday, the 29th, its audio generation model in artificial intelligence (AI) that copies human voices. The company said, however, that it is choosing to just show the technology to the public, without launching it “widely” at this time, given the risks of misuse.
The Voice Engine tool can recreate the voice of a specific person and make them read messages written by the user, based on a recording of just 15 seconds of the original speaker, according to the OpenAI statement. The company said it developed the technology in late 2022, and has begun preliminary testing it with a small group of “trusted partners.”
In this Friday’s announcement, OpenAI shared insights into applications so far with the rationale of starting a dialogue about the responsible adoption of what it called “synthetic voices.” “According to the conversations and the results of small tests, we will make a more informed decision about how or if we will use this technology”, he said, in a post on his blog.
OpenAI said it has identified ways to apply it “for good,” such as: providing reading assistance to people who can’t read; translate content; help patients recover their voice; and other therapeutic uses.
On the other hand, the company recognized that the technology poses serious risks, especially in election years around the world. The company said it hopes that the Voice Engine will motivate society to build greater resilience against challenges. She proposes, for example, gradually eliminating voice authentication for accessing bank accounts, exploring policies to protect the use of individuals’ voices, and accelerating the development of techniques for identifying content created by AI.
2024-03-29 22:50:00
#OpenAI #introduces #technology #clones #human #voices #chooses #release
