Sample Encoding

The result of the digital sampling process is a sequence of single samples. One sample is a digital representation of a signal's value at a certain time.

The value of a sample can be interpreted and encoded in several ways. The simplest one is linear encoding. This means that each sample's value directly represents the analogue signal's value multiplied with a constant factor. This is easy to handle, but has the disadvantage that noise will be audible especially on low amplitudes, where it disturbs most, and less audible on high amplitudes, where it is less audible.

One way to reduce the influence of noise is non-linear encoding. This means that lower amplitudes are amplified before processing. As lower amplitudes are amplified, their distance from noise increases and the quality improves. The most common methods for this are A-Law and U-Law encoding - some standardized logarithmic amplification curves, used in digital telephony (ITU G.711 standard).

Nonlinear Encoding

Nonlinear Encoding