TECHNOLOGICAL ACHIEVEMENTS Photonic Sensors Computer-Aided Structural Design Artificial Vision Virtual Therapy FalconView Near-Field Antenna Measurements Millimeter Wave Radar Near-Field Sampling Applied Chaos Three-Dimensional Vision Solar Energy Fused Silica Radomes Pulse Combustion Monodisperse Aerosol Generation Digital Signal Processing BASIC DISCOVERIES TABLE of CONTENTS

Digital Signal Processing

Courtesy ASP Inc. Georgia Tech research by Drs. Ronald Schafer and Thomas Barnwell III has improved the technology for digital transmission and storage of speech signals and led to a new national standard in digital speech compression. Click here to see larger 70k version.

Work at Georgia Tech to improve the technology for digital transmission and storage of speech signals led to a new national standard in digital speech compression. The U.S. Department of Defense's Digital Voice Processors Consortium awarded the designation in May 1996 following a three-year, federally sponsored competition among a host of highly respected research labs.

The digital speech standard follows years of research at Georgia Tech in general speech and audio processing. The work includes three-dimensional modeling of speech production, low-rate speech coding, speech analysis and synthesis, speaker characterization, speech quality measurement, automatic speech recognition and digital audio coding at high quality.

This new standard is improving telecommunications technology and reverberating throughout the cellular, Internet, telephone and wireless communications industries.

The standard resulted from research started in the late 1980s by graduate student Alan McCree and Dr. Thomas Barnwell III, a professor in the School of Electrical and Computer Engineering. In essence, the work involves development of processing algorithms that incorporate knowledge of speech production processes and the intrinsic redundancies of human speech.

With those algorithms, the computer-sampled voice signal can be stripped down in the laboratory and then reconstituted without the listener perceiving any great difference.

Researchers have developed a voice coder or "vocoder" that systematically constructs high-quality synthetic speech by "remembering" and "using" voice redundancies. Key to the vocoder's performance is a filter through which the excitation source passes.

Ten parameters in the filter interact with the source in such a way that the natural voice signal it was derived from can be defined, then reproduced.

Barnwell is co-founder and president of Atlanta Signal Processors Inc., which sells hardware and software tools for digital signal-processing algorithm development and multimedia on high-speed microprocessors.

For more information, contact Dr. Thomas Barnwell III, School of Electrical and Computer Engineering, Georgia Tech, Atlanta, GA 30332-0250. (Telephone: 404-894-2914) (E-mail: tom.barnwell@ee.gatech.edu)

Contents    Research Horizons    GT Research News    GTRI    Georgia Tech

Send questions and comments regarding these pages to Webmaster@gtri.gatech.edu

Last updated: October 25, 1999