MOVES · MOdelling Vocal Expression in Schizophrenia

The human voice is a powerful tool for social communication. In recent years, Artificial Intelligence (AI) fostered the development of advanced voice systems, able to infer considerable information from the speaker’s voice, such as emotional and mental states, mood information and personality traits. Individuals with schizophrenia (SZ) tend to present voice atypicalities, which are related to core clinical symptoms and social impairment. Recent advances in voice technology may lead the way to a revolution in the study of voice disorders. They may allow to disentangle the affective, cognitive and social mechanisms responsible for voice atypicalities, assist clinicians in diagnosis and monitoring of the disorders, and enhance their capability to capture the complex relationship between vocal behaviour, emotion regulation and clinical features. However, our present understanding of voice abnormalities in SZ is very poor, limited by the lack of comprehensive models and systematic approaches to study voice production.MOVES aims at providing a solid understanding of the implications of atypical voice patterns in SZ: through the application of machine learning and signal processing technologies (AI), I will provide a first comprehensive account of the mechanisms underlying voice atypicalities, assess their impact on clinical evaluations, and create the foundations for more reliable and evidence-based screening tools. The project aims to foster multi-centric and international collaborations to overcome important limits of this research field, such as the need for cross-linguistic studies, larger datasets, and open and collaborative research. MOVES pioneers a new area of research at the intersection between cognitive neuroscience, psychiatry, computational science and AI. An innovative aspect of the project is the intention to translate recent AI technological advances into clinical settings, to improve the way we conceptualise, assess and monitor voice disorders in SZ.