MIT EECS — DENSO Undergraduate Research and Innovation Scholar
Development of Automatic Speech Analysis System
The Speech Communication Group is working towards completion of a speech analysis system modeled on human speech understanding, based on the detection of acoustic cues, called landmarks. The Speech Communication Group’s child speech project involves analyzing speech from typically and atypically developing children. This analysis is important in potentially diagnosing clinical disorders. The group is working on automating the process of landmark labeling, through the development of modules, each identifying specific landmarks. The landmark labels are used to create diagnostic outputs, like landmark production profiles and decision tree analysis of speech modifications. The SuperUROP project involves the development of the remaining modules and the integration of the entire system.
The Speech Communication Group is a multidisciplinary laboratory, combining electrical engineering and computer science with linguistic theory and cognitive science. I have the opportunity to learn a lot, in and outside my field. This research project will allow me to apply my knowledge from artificial intelligence and signals and systems classes, and to learn more about linguistic theory and cognitive science.