Speech and Audio Signal Processing: Processing and Perception of Speech and Music, Second Edition

CHAPTER 32

VOCODERS

32.1 INTRODUCTION

We are now in a position to define complete vocoder systems, often called analysis–synthesis systems. Detailed descriptions of spectral envelope estimations employing three analysis methods were outlined in Chapters 19, 20, and 21. These methods are filter banks, cepstral analysis, and linear prediction. In addition, speech synthesis was treated in Chapter 30 and pitch detection was treated in Chapter 31. In this chapter, we combine this material into complete analysis–synthesis structures. The following chapters will elaborate on specific forms of vocoders.

The primary application of vocoder systems is source coding to reduce the required rate of transmission. Source coding is also useful for efficient storage of speech and music, for example, in voice answer-back systems or in compact disks. As Dudley pointed out (see the preface to Chapter 3), source coding is useful in secrecy systems; it is also useful as an aide to help us understand how speech is produced, which in turn may be helpful for exploring diagnostic tools in speech and hearing pathologies. Another application is speech transformations, which we discuss in Chapter 40.

32.2 STANDARDS FOR DIGITAL SPEECH CODING

Figure 32.1 shows the standardized bandwidths and coding rates for four speech and music applications. The purpose of source coding research is to devise methods of lowering ...