Out in Nature Portfolio SciReports today, our work on detecting AI-powered voice clones. TL;DR: AI-generated voices are nearly indistinguishable from human voices. Listeners could only spot a 'fake' 60% of the time. By comparison, randomly guessing gets 50% accuracy. Furthermore, listeners thought side-by-side audios of real and AI-cloned voices were the same identity 80% of the time. With great thanks to Hany Farid and Emily Cooper for wonderful co-authorship as always, and ElevenLabs for supporting this work.
