Voice Identity

Voice identity research focuses on understanding how voices are recognized and how this information can be manipulated or protected. Current research emphasizes developing robust models, often leveraging self-supervised learning and techniques like normalizing flows, to improve speaker identification and voice anonymization, particularly for challenging scenarios such as singing voices and cross-lingual speech. These advancements have significant implications for applications ranging from improved speech synthesis and personalized voice assistants to enhancing privacy in voice-based systems and forensic speaker recognition.