Electrical and Computer Engineering
Center for Multimedia Communication
System and Applications R&D Center
Texas Instruments Leadership University Seminar
Monday, October 28, 2013
"Sensor and Audio Analytics in TI R&D”
to 5:00 PM
A227 Abercrombie Engineering Laboratory
6100 Main St
Houston, Texas, USA
The Audio and Sensor Analytics branch in the Embedded Processing Systems Lab at TI works on projects that extract information from audio, inertial and other sensors. In this talk, Devangi will give a brief overview of these projects, and then discuss in more detail her work on Low-MIPS Audio. To make the abundant sensors that exist around us “smart”, an always-on event detection system has garnered interest. An always-on event detection system requires both an ultra-low-power device either analog or digital, and low-MIPS algorithm that can detect and classify the events. This event may be in the form of audio or speech signals. Devangi will talk about one such application of voice wake-up that uses a keyword recognition algorithm, which is targeted towards an ultra-low-power microcontroller.
Host: Behnaam Aazhang
Biography of Devanghi Parikh:
Devangi Parikh received her M.S. and Ph.D degrees from Georgia Institute of Technology in 2008 and 2012 respectively. She joined the Audio and Sensor Analytics branch within the Embedded Processing Systems Lab at TI in 2012 after interning with the same group for three summers. As part of her thesis work, she worked on single- and multi-microphone speech enhancement algorithms. At TI, she is investigating the development and implementation of low-MIPS audio algorithms.