ABSTRACT
Equipped with the basic principles and techniques discussed in Chapter 1 for analyzing general discrete-time signals and systems, we are now in a position to apply them to analyzing sampled or discrete-time speech signals. Most speech processing applications utilize certain properties of speech signals in accomplishing their tasks. This chapter describes these properties or features and how to obtain them from a speech signal s(n), i.e., speech analysis. This typically requires a transformation of s(n) into a set of parameters, or more generally into a set of signals, with the purpose often being data reduction. The relevant information in speech can often be represented very efficiently. Speech analysis extracts features which are pertinent for different applications, while removing irrelevant aspects of the speech.
