Post by Ewha Womans University 이화여자대학교

Can we stop AI from mimicking a voice without your consent? 🎙️🚫   While zero-shot TTS offers incredible expressivity, it also poses serious risks of voice misuse. Professor Jiyoung Lee and her team at the department of AI introduce TruS—a paradigm-shifting framework that erases specific speaker identities without the need for costly retraining. By steering identity-specific activations during inference, TruS effectively prevents the synthesis of both seen and unseen 'opt-out' speakers while keeping prosody and emotion intact. A scalable safeguard for the future of speech synthesis.   Presented at International Conference on Acoustics, Speech, and Signal Processing (ICASSP) May 2026 Read more: https://lnkd.in/gvvcCG3G #EwhaResearch #ICASSP2026 #SpeechAI #MachineUnlearning #AISafety #Privacy #ZeroShotTTS #AIEthics