Anonymous Attendee       11:56 AM
How is classifying emotions from acoustic signals helpful in predicting OCD behaviour?

sneha das       12:02 PM
Its shown that OCD, and in general mental disorders have a lot of influence of audio and speech. For instance, a depressed person shows a lot less variability in tone and pitch of their speech. One prominant symtyom of OCD is anxiety, and axiety can also be detected from spectral flux, pitch, loudness etc. So we hope speech, along with the other signal modalities will help understanding OCD progression and severity during therapy sessions.


Anonymous Attendee       11:56 AM
Interesting project. You pose one challenge as: generalising speech patterns.  How would this be achieved across languages and place-based dialects that differ and are highly nuanced, in terms of emotional meaning?

sneha das       12:09 PM
Good question! We see generalization as a stumbilng block in terms of universal models and accuracy numbes. But then the question is do we really want to generalize over all population, dialects, corepora, etc. Because we can already see that emotions in speech may not always be universal. So, precision psychitry could be a potential approach to explore for this work.