Speaker Age

Speaker age estimation and verification from speech are active research areas aiming to accurately determine a person's age from their voice, either as a single value or to account for age variations in speaker recognition. Current research focuses on developing robust models, often employing techniques like label distribution learning and adversarial learning, to address the inherent ambiguity and variability in vocal characteristics across ages and to disentangle age-related features from speaker identity. These advancements have implications for improving automatic speech recognition systems for elderly or dysarthric speakers and enhancing security applications like speaker verification across age ranges.