What kind of method is ADPCM?
Descriptive Answer:
| » | DM and ADM methods |
For the ADM (Adaptive Delta Modulation) method, the quantized width (
) is adjusted according to the rapid changes to improve the response of the voice waveform. Encoding one bit is based on this means. If a value of 1 or 0 continues for a certain period, the quantized width (
) is enlarged for quicker response. Figure 10.1 (b) illustrates such quick response.
| » | ADPCM method |
For example, in the case of four-bit ADPCM, the upper one bit stands for polarity (increase or decrease) data, while the lower three bits determine the multiplier factor to the basic width of quantization (
). (The
value depends on the correlation with the past data.) This means that about 400 pieces of data [(about 50
pieces) x 8 (for 3 bits)] can be changed at a time. Therefore, approximately 400 pieces of data are available. Figure 10.2 illustrates this case.
Furthermore, in the three-bit ADPCM, the upper one bit stands for polarity (increase or decrease) data, while the lower two bits determine the multiplier factor to the basic width of quantization (
). This means that about 200 pieces of data [(about 50
pieces) x 4 (for 2 bits)] can be changed at a time. Therefore, 200 pieces of data are available.
The ADPCM method allows the achievement of high-quality sound in a relatively simple configuration, and the easy creation of voice data.
Note, however, that there is no compatibility between ADPCM data created by different manufacturers.
(Note) When playback starts from part of a phrase, normal waveforms cannot be reproduced because the ADPCM algorithm is employed. See Q10 and A10.
| Figure 10.1 DM and ADM Methods | ![]() |
|---|---|
![]() |
|
| Figure 10.2 4-bit ADPCM | ![]() |



