An adaptive quantizer for DPCM(differential PCM) is designed, and various performance parameters are compared with the calculated or simulated results. For varieties in experiment, the quantizer is implemented to be a programmable type.
Some typical multiplier sets $\{Mi\}=\{\Gamma^{mi}\}$ are simulated for a stationary Gaussion Signal, and the characteristics of $\{mi\}$, $\Gamma$are observed. The performance of the ADPCM coder is checked for real speech signal, where the measured parameters are mean and variance of the quantizer step-size, signal-to-quantizing noise ratio, and signal spectrum.
It is shown that the static performance and adaptation speed of the quantizer are strongly affected by $\{mi\}$ and $\Gamma$, respectively. The signal spectrum is observed as an indirect measure for quantizer performance and signal quality.
