Speech and Speaker Modeling Group
Research Projects
Dialect / Accent Classification This project is focused on developing new algorithms for automatic detection and classification of Dialect/Accent in natural speech communication.
In-Set Speaker Recognition Here, our studies are focused on small enrollment and test data sizes (2-8 sec), with open domain content.
Speaker Normalization In Normalization, we focus on making speech systems robust under differing operational conditions.These include variation in environment, language, bandwidth, noise, and speaker differences.
SpeechFind A spoken document retrieval system serving as the search engine for the National Gallery of the Spoken Word and Collaborativ e Digitization Program. This project supports automatic online search of multi-media content.
UTDrive UTDrive is part of a multi-year NEDO-supported international collaboration between Japan, Italy, Singapore, Turkey, and USA.
Speech Under Stress This project deals with finding the features which help identify stress in speech. The focus is on nonlinear and linear features for detection and assessment of stress (cognitive and physical), Lombard effect, and emotion in speech.
Whispered Speech The goal of this project is to develop automatic algorithms to detect whispered, soft, loud, and shouted portions of speech, and to contribute to improved overall speech systems.
Center for Robust Speech Systems
CRSS Office :
972 883 4749
e-mail: Speech
Prof. John H.L. Hansen :
972 883 2910
e-mail: John H.L. Hansen
Updated: March 27, 2008