From Future plc: 2024-03-30 11:51:49

OpenAI has introduced Voice Engine, a tool that mimics a person’s voice after just 15 seconds of recorded audio. This model learns how a person speaks from a short clip and can then generate speech with realistic emotions. OpenAI recognizes the potential for misuse and is cautious about a wider release.

The company plans to engage in discussions with various organizations to gather feedback on Voice Engine. Testing is currently underway with parties agreeing not to impersonate others and to disclose when the voice is AI-generated. OpenAI has also implemented watermarking to indicate when the voice is synthetic.

OpenAI emphasizes the importance of voice authentication and prevention of voices similar to famous figures. The future of Voice Engine remains uncertain, with decisions on public release pending further feedback. The company stresses the need for global awareness of the technology’s potential impact, regardless of its eventual deployment.



Read more at Future plc: OpenAI’s latest tool can recreate human voices in just 15 seconds — meet Voice Engine