Smarter devices will understand emotion, protect confidentiality

Using the insightful nature of speech and vocal patterns, a Texas A&M University engineer is developing a digital program that monitors and tracks a user’s emotional state while keeping their identity completely anonymous. Her research has the potential to transform everyday devices into valuable assets for psychological healthcare and future research.

Every voice contains unique attributes (such as timber and pitch) that can reveal such things as age and gender. This information can then be used to identify a speaker, which is especially concerning when personal health information is involved.

“What we want to do is make ordinary devices be able to understand emotion, while erasing any information related to the identity of the speaker,” said Theodora Chaspari, an assistant professor in the Department of Computer Science and Engineering. “So we are transforming the speech signal and deriving measures that are emotion-dependent, behavior-dependent, but not identity-dependent.”

By removing the link between voice and speaker to provide anonymity to users, Chaspari’s emotion monitoring program is vital in the development of emotional healthcare and research in a digital world.

Her project is funded by a 2019 grant from the Program to Enhance Scholarly and Creative Activities (PESCA), funded by the Texas A&M Division of Research.