Databases Reference
In-Depth Information
11.5.1 Adaptive Quantization in DPCM
In forward adaptive quantization, the input is divided into blocks. The quantizer parameters are
estimated for each block. These parameters are transmitted to the receiver as side information.
In DPCM, the quantizer is in a feedback loop, which means that the input to the quantizer is not
conveniently available in a form that can be used for forward adaptive quantization. Therefore,
most DPCM systems use backward adaptive quantization.
The backward adaptive quantization used in DPCM systems is basically a variation of the
backward adaptive Jayant quantizer described in Chapter 9. In Chapter 9, the Jayant algorithm
was used to adapt the quantizer to a stationary input. In DPCM, the algorithm is used to adapt
the quantizer to the local behavior of nonstationary inputs. Consider the speech segment shown
in Figure
11.7
and the residual sequence shown in Figure
11.8
. Obviously, the quantizer used
around the 3000th sample should not be the same quantizer that was used around the 1000th
sample. The Jayant algorithm provides an effective approach to adapting the quantizer to the
variations in the input characteristics.
Example11.5.1:
Let's encode the speech sample shown in Figure
11.7
using a DPCM system with a backward
adaptive quantizer. We will use a third-order predictor and an eight-level quantizer. We will
also use the following multipliers [
124
]:
M
0
=
0
.
90
M
1
=
0
.
90
M
2
=
1
.
25
M
3
=
1
.
75
The results are shown in Figure
11.10
. Notice the region at the beginning of the speech
sample and between the 3000th and 3500th sample, where the DPCM system with the fixed
quantizer had problems. Because the step size of the adaptive quantizer can become quite
small, these regions have been nicely reproduced. However, right after this region, the speech
output has a larger spike than the reconstructed waveform. This is an indication that the
quantizer is not expanding rapidly enough. This can be remedied by increasing the value of
M
3
. The program used to generate this example is
dpcm_aqb
. You can use this program to
study the behavior of the system for different configurations.
11.5.2 Adaptive Prediction in DPCM
The equations used to obtain the predictor coefficients were derived based on the assumption
of stationarity. However, we see from Figure
11.7
that this assumption is not true. In the
speech segment shown in Figure
11.7
, different segments have different characteristics. This
is true for most sources we deal with; while the source output may be locally stationary over
any significant length of the output, the statistics may vary considerably. In this situation, it
is better to adapt the predictor to match the local statistics. This adaptation can be forward
adaptive or backward adaptive.
