Diego

CHAPTER 8 Speech Coding and Channel Coding
8.1 Introduction
In Chapter 7 we introduced speech and channel coding with reference to the GSM. In this chapter we provide detailed descriptions of speech and channel coding schemes. We first discuss speech coding methods and attributes of a speech codec. (The speech codec is also called a voice codec or vocoder. It is a hardware circuit that convertsthe spoken word into digital code and vice versa.) These are then followed by a brief discussion of linear-prediction-based analysis-by-synthesis (LPAS) method. We also discuss the QCELP, EVRC, EFR, and AMR codecs that are used in WWAN systems. We then focus on channel coding, concentrating on three channel coding schemes — convolutional code, Reed-Solmon (R-S) block code, and turbo code. Theconvolutional codes have been used in direct sequence spread spectrum CDMA (IS-95) and R-S and turbo code are the proposed channel coding schemes for cdma2000 and WCDMA. To achieve reliable communication with minimum possible signal power is the main objective of a communications system engineer. Speech coding is used to save bandwidth and improve bandwidth efficiency whereas channel coding is employedto improve signal quality and reduce bit-error-rate (BER). The idea of using a channel coding scheme is to recover from errors that occur during transmission over the communication channel. The channel coding strategy is aimed at allowing the transmitter to use minimum possible signal power in accomplishing the design objective of providing a specified error rate.

8.2 Speech Coding
The speechquality of codecs operating at a fixed bit rate is largely determined by the worst-case speech segments, i.e., those that are the most difficult to code at the given rate. Variable rate coding can provide a given level of speech quality at an average bit rate that is substantially less than the bit rate that would be required by an equivalent quality fixed rate codec [1–4]. The original CDMA codecknown as Qualcomm code-excited linear predictive (QCELP), IS 96A was an 8 kbps code-excited linear predictive (CELP) variable rate

215

216

8

Speech Coding and Channel Coding

codec designed for use in the 900 MHz digital cellular band [11,12]. The desire for improved voice quality spurred the TIA to begin working on a 13 kbps CELP variable rate codec to provide higher quality voicetransmissions. The 13 kbps CDMA codec takes advantage of the higher data rate (14.4 kbps as compared to 9.6 kbps for the 8 kbps codec) to improve speech quality. It has produced mean opinion scores (MOS) close to toll-quality voice, the benchmark used for comparison. Unfortunately, system capacity and cell coverage are reduced by the higher data rate codec.

8.2.1 Speech Coding Methods Speechcoding is the process for reducing the bit rate of digital speech representation for transmission or storage, while maintaining a speech quality that is acceptable for the application. Speech coding methods are classified as waveform coding, source coding, and hybrid coding. The following sections will explain these concepts. In Figure 8.1, the bit rate is plotted on a logarithmic axis versus speechquality classes of “poor to excellent” corresponding to the five-point MOS scale values of 1 to 5, defined by the International Telecommunications Union (ITU). It may be noted that for low complexity and low delay, a bit rate of 32 to 64 kbps is required. This suggests the use of waveform codecs. However, for low bit rate of 4 to 8 kbps, hybrid codecs should be used (see Figure 8.1). These types ofcodecs tend to be complex with high delay [5–8].

SPEECH QUALITY WAVEFORM CODECS HYBRID CODECS

EXCELLENT

GOOD

FAIR

POOR VOCODERS BAD 1 2 4 8 16 BIT RATE (kbits/s) 32 64

Figure 8.1

Quality of service versus bit rate.

8.2

Speech Coding

217

8.2.2 Speech Codec Attributes Speech quality as produced by a codec is a function of transmission bit rate, complexity, delay,...