Quantisation is a process by which the infinite range of values in a continuous interval is represented by a finite number of discrete values.
Quantisation is necessary in digital systems as each sample must be represented by a binary number with a finite number of bits. The number of bits per sample, known as the bit depth, is typically 16 or 24 in digital audio. An n-bits binary number can represent 2n distinct values. So a 16-bit binary number can represent 216 = 65,536 values, while a 24 bit binary number can represent 224 = 16,777,216 values.
The level of each sample must be represented by the available binary value that is closest to it. As a result there is often a small difference between the actual level of the sample and its binary representation. This difference is called quantisation noise. The greater the bit depth of the digital representation, the greater the number of binary level available and therefore the lower is the quantisation noise. This is why greater bit depths are preferred in digital audio.
©Wayne Butcher