What is quantization


Quantization is part of the digitization of analog signals. For this purpose, the analog signals are sampled at regular intervals at the sampling rate and their voltage value is converted into a digital value at each of these sampling times. The analog signal can only be expressed in a finite number of digital values.

This conversion into numbers and measurable quantities is called quantization. For example, an analog value can be represented with a sample depth of 4 bits, 8 bits, 16 bits, 24 bits or 32 bits, which corresponds to a quantization of 16, 256, 65,536, 16,777,216 or 4,295 x 10exp9 different levels. The higher the quantization, the greater the dynamic range and the lower the possible percentage quantization error in relation to the original signal.

Quantization methods work with linear and non-linear conversion. With linear quantization, the quantization interval is constant and therefore independent of the signal level. This leads to the disadvantage that the signal-to-noise ratio (SNR) is worse at low signal levels than at large ones.

It is different with non-linear quantization. With this method, the input signal is pre-distorted, small levels are raised, large levels are lowered. Non-linear quantizations work with non-linear characteristics that have logarithmic functions or are designed as segment characteristics. The 13-segment characteristic, which consists of 13 linear segments, is known. With non-linear quantization, the input signals are first compressed, then quantized, then transmitted and expanded again. This method significantly improves the signal-to-noise ratio and the dynamic range.