Ina Kodrasi, Idiap Research Institute

Title: Advances in signal processing and machine learning for clinical speech assessment

Abstract: With growing population numbers and an aging demographic, the prevalence of speech disorders is steadily increasing. These disorders often arise from neurological conditions such as Parkinson’s disease, Amyotrophic Lateral Sclerosis, or stroke, but may also result from impaired auditory feedback or delayed language acquisition in individuals with hearing loss. Diagnosing and differentiating among speech disorders, as well as monitoring their progression, typically requires time-consuming and costly auditory-perceptual evaluations conducted by speech-language pathologists. To alleviate this burden on the healthcare system, current research is focused on developing automatic speech processing technologies. This talk will present state-of-the-art signal processing and machine learning approaches for characterizing and diagnosing various speech disorders. It will also address the technological challenges individuals with speech disorders face when interacting with voice-controlled systems, underscoring the importance of inclusive and accessible technology design.