Sampling

How do we convert an analog audio signal into a digital one?

The process is known as sampling and is done like this:

Many times a second (the sampling rate) look how big the signal is and give it a value, e.g. from 0 to 256, that can be represented by a binary code.

This will produce lots of digital information. If the sampling rate is 16kHz, i.e. 16,000 samples every second, and we assign 8 bits to every sample then we produce 16,000 x 8 = 128,000 bits per second of information.

The electronic system that receives this information can then, if needed, convert this information back into an analog form by effectively "joining the dots".

There is a compromise between the quality of the signal and the amount of information it consists of.

If we sample at a very high rate, e.g. 44kHz, and assign a lot of bits per sample then the quality of the signal is very high but it will take a long time to transmit and, if stored, will take up a lot of memory space.

If we sample at a much lower rate then we lose quality. You should see that we will lose the high frequency components of the signal. But there are less bits per second involved and we may even be able to "stream" the signal i.e. send it in real time without the need to store it first.

Here is a useful comparison;

To reproduce a signal with a frequency of 20kHz faithfully we need a sampling rate of 40kHz, twice as much. As 20kHz is the limit of human hearing then there is not much point sampling music at rates of above 40kHz.

The minimum sampling rate required is called the Nyquist Rate.

It is equal to twice the highest frequency contained within the signal.