From Crypto Times: 2024-03-30 12:11:14

OpenAI is showcasing a new feature that can convincingly narrate text with a human-like voice, raising concerns about deepfake risks. The Voice Engine technology has been shared with ten developers, following input from stakeholders to limit early access. The company is collaborating with various sectors to address potential risks, especially during an election year.

Voice Engine by OpenAI can replicate specific voices with just 15 seconds of recorded audio, unlike previous attempts at generating fake voices. The technology has shown impressive quality but also requires cautious use due to safety concerns. Partners like a neuroscience institute use Voice Engine to aid patients in voice recovery, demonstrating its potential benefits.

OpenAI’s speech model is being used to translate audio into different languages for podcast translations and educational content. Partners in the testing program must follow strict usage policies, obtain consent, and disclose AI-generated voices to listeners. OpenAI is seeking feedback from experts for a global understanding of the technology’s impact and resilience in society against advanced AI technology.



Read more at Crypto Times: OpenAI Preview New Voice Mimicry and Text Reading Tool