Digital Speech: Coding for Low Bit Rate Communication Systems, 2nd Edition

3

Sampling and Quantization

3.1 Introduction

In digital communication systems, signal processing tools require the input source to be digitized before being processed through various stages of the network. The digitization process consists of two main stages: sampling the signal and converting the sampled amplitudes into binary (digital) codewords. The difference between the original analogue amplitudes and the digitized ones depend on the number of bits used in the conversion. A 16 bit analogue to digital converter is usually used to sample and digitize the input analogue speech signal. Having digitized the input speech, the speech coding algorithms are used to compress the resultant bit rate where various quantizers are used. In this chapter, after a brief review of the sampling process, quantizers which are used in speech coders are discussed.

3.2 Sampling

As stated above, the digital conversion process can be split into sampling, which discretizes the continuous time, and quantization, which reduces the infinite range of the sampled amplitudes to a finite set of possibilities. The sampled waveform can be represented by,

where s_a is the analogue waveform, n is the integer sample number and T is the sampling time (the time difference between any two adjacent samples, which is determined by the bandwidth or the highest frequency in the input signal).

The sampling theorem states ...