Diagnostic Information
Clinical Applications

Features extraction and information processing in audition

Previous and currrent work focus on identifying specific regions of the brain (STS) selective to human voices. Using sex gender as a tool, we are currently investigating if categorization mechanisms are similar for human voice and other sounds. For instance, from a behavioural viewpoint, we showed that gender categorization differs from the categorization of musical instruments. In addition, while it has been recognize that pitch cannot be used for gender categorization because of the large overlap between male and female voices, we demonstrated that, when the timber was ambiguous, the brain does use the pitch. Using a similar approach, we are testing if gender categorization differs from phoneme categorization using identical stimuli. This work, is also completed by testing stroke patients in order to identify possible key regions unperpinning those mechanisms.

Mathematical modeling of voice gender categorization. We modeled the categorization performance of 30 subjects (B) as an interaction between pitch and gender. Results suggest that while easy male and female voices are categorized based on timbre information, the pitch is used to reach a decision when stimuli are ambiguous (E, F).