1. Digital Audio

Digital Audio (slides 3–4)

Two characteristics of digital sound – sampling and quantisation – permit to introduce en passant the main concepts of digitisation.

Let’s consider an analogue signal, a sound in the real world:

  • sampling is the reduction of this continuous signal to a discrete signal (to a sequence of tiny sections);
  • quantisation is the process of mapping an infinite set of input values to a finite set of output values (to rounded values at a fixed precision).
Sampling (slide 5)

Sampling measures how close-knit the digitisation is done. In a certain sense, it measures the sound quantity,1 and is similar to the resolution for the image. If the frequency increases, then the number of different values that are generated can increase. The unit is hertz (Hz), the number of measurements taken during every single second.

Some current sampling rates are:

  • 44.1 kHz is the CD quality;
  • 48 kHz is a sampling rate often used for postproduction. It can also be chosen instead of 44.1 kHz for access and diffusion;
  • 96 kHz is a good sampling rate for archiving;
  • 192 kHz is an excellent sampling rate for archiving, but it needs the double storage space in comparison to 96 kHz.

In the real world the archive must made the best possible compromise in a longterm perspective: What sampling can the archive afford over a long period of time, at least over many decades?

Quantisation (slide 6)

Quantisation measures how precise the digitisation of one cycle of sampling is done. In other words, it tells how many different steps can be differentiated at every measurement. This is one of the factors that determine the sound quality.

Typical sound quantisation rates are:

  • 16 bit for access and diffusion;
  • 24 bit for archiving.

This switch from 16 bit to 24 bit gives 8 additional bits or 1 additional byte. In a binary system every bit has one of the two values 0 or 1. That means that these 8 additional bits give 2 power 8 equals 256 times more possibilities to code the sound quality. Otherwise said, by increasing from 16 bit to 24 bit the quantisation, the needed storage space increases by 50%, yet theoretically the sound quality increases by 25 600%.


This designation by the author is not used in the technical literature.