Paper Authors

Abstract

NOTE: The first page of text has been automatically extracted and included below in lieu of an abstract

SESSION 2520

REAL-TIME DEMONSTRATIONS OF QUANTIZATION AND PREDICTION USING THE ‘c31 DSK

J.W. Goode, S.A. McClellan, IBM Corporation UAB Elec. & Comp. Engr

Abstract

In the study of data compression, signal processing, and digital communications (among others), the topics of quantization and linear prediction play a central role. For example, the finite wordlengths required for storage of digital filter coefficients as well as the number of bits used in “acquiring” a signal can be of significant concern in many applications. The effect of quantization in these cases is often quite difficult to explain in a fashion which makes sense to undergraduate students. Since much of the theory surrounding quantizer optimization and optimal linear prediction is well-beyond the undergraduate level, and since a good understanding of these basic topics is increasingly important in the “information age”, we have approached the problem from the perspective that “a good demonstration is worth a thousand equations”. To enable our demonstrations, we use an inexpensive TI ‘c31-based DSK. The DSK is quite portable, flexible, has excellent analog input/output capabilities, and has more than adequate horsepower for our purposes. Using some assembly language and some C-language constructs, we have implemented a full- featured quantization and prediction platform for real-time acquisition, processing, and reconstruction of speech. We have found that the real-time demonstration of both quantization and prediction in a carefully selected progression is very helpful to undergraduate (as well as graduate) students in grasping these sometimes difficult concepts.

I. INTRODUCTION

Digital communications systems have been used for many years to transmit analog signals because of inherent advantages that digital encoding presents in comparison to analog transmission. Some advantages of digital communication include low susceptibility to transmission noise, applicability of digital encryption, and an integration of various types of data (e.g., computer data, digitized speech and images) in a common transmission format. In the study of digital communications as well as data compression and signal processing, the topics of quantization and linear prediction play a central role. Unfortunately, the effect of quantization is often quite difficult to explain in a fashion which makes sense to undergraduate students. Since much of the theory surrounding quantizer optimization and optimal linear prediction is well-beyond the undergraduate level, we have approached the problem from the perspective that “a good demonstration is worth a thousand equations”.

To enable our demonstrations, we focus on the well-understood mechanism of speech production, and we utilize the inexpensive and quite flexible “DSP starter kit” from Texas Instruments. The starter kit is comprised of an evaluation board for the TMS320C31 50 MHz digital signal processor (DSP) which has the necessary analog/digital conversion features as well as an interface for communications with the

• Citation

• Format
  • APA
  • APA - LaTeX bibitem
  • MLA
  • MLA - LaTeX bibitem
  • Bibtex
  • EndNote - RIS

McClellan, S., & Goode, J. (1998, June), Real Time Demonstrations Of Quantization And Prediction Using The 'c31 Dsk Paper presented at 1998 Annual Conference, Seattle, Washington. 10.18260/1-2--7383