Coding Quality of MPEG-4 Natural Speech Coding Tools
Coding quality of CELP [15] is depicted in Fig.
16 (a) and (b). The narrowband mode with MPE, which supports multiple bitrates
with a single algorithm, exhibits comparable quality to those of other standards
( ITU-T G.723.1, G.729, ETSI GSM-EFR) optimized at a single bitrate. The wideband
mode with MPE at 17.9 kbit/s provides comparable quality to those of ITU-T
G.722 at 56 kbit/s and MPEG-2 Layer III at 24 kbit/s.
Speech quality by scalable coding naturally is degraded compared to nonscalable
coding because of split loss of the bitstream. However, MPEG-4 CELP is successful
in minimizing the difference in speech quality between scalable and nonscalable
coding thanks to its algorithmic features. Fig. 16 (c) exhibits the coding
quality of HVXC [15]. MPEG-4/HVXC clearly outperforms
the reference, FS1016 [22] at 4.8 kbit/s (FS1016
is a US Department of Defense standard that is most commonly used as a reference
at a bitrate lower than 5 kbit/s).
Figure 16: Coding Quality of MPEG-4 Natural Speech Coding Tools
