Speech processing using the TS32010 - a case study

Abstract

This case study illustrates the now classical Linear Predictive Coding (LPC) method of speech compression used to reduce the bit rate in digital speech transmission systems. The chapter begins with a non-rigerous introduction to the theory of linear predictive coding. This is followed by an explanation of the real-time algorithms used to calculate the parameters required to synthesise individual pitches of voiced speech. The study ends with a description of how these algorithms are implemented on the TMS32010, the problems incurred and results obtained.