Often in high-end audio discussions someone will praise the virtues of analog audio, usually vinyl records. Soon enough someone else will claim that digital audio, especially high-resolution audio, is best. But what is this ‘analog’ and ‘digital’ stuff anyway? Let’s try to explain, without getting too technical and while steering clear of any ‘religious’ wars.
Analog
An analog signal is one which is analogous, or similar in form, to the original source that produced it. What do we mean by similar? In analog audio the pattern of the audio signal is similar to the pattern of air pressure fluctuations of the original sound. The audio signal level is high at an instant corresponding to an instant when the air pressure was high. The frequency of the audio signal also matches to the frequency of the original sound at the corresponding time.
The most common analog audio sources are vinyl records and tapes. Let’s use a vinyl as an example. A record has a fine spiral groove which the tiny stylus (needle) of the phono cartridge follows as the record is played.
The music is encoded in the wavy pattern of the record groove, which follows the same variation as the original sound. The phono cartridge detects the pattern of the record groove and converts it into a very weak electrical signal, which itself follows the variations of the original sound. Your sound system amplifies this weak signal and uses it to drive your loudspeakers so they produce a pattern of air pressure fluctuations (sound) which is, hopefully, identical to the originally recorded sound.
Digital
A digital signal is a sequence of numbers which represents the original source signal. The source is sampled at regular intervals and the value at each sampling instant is converted to a number which is stored. This process is called analog to digital conversion.
The sequence of stored numbers is the digital representation of the original source signal.
In your digital audio system a digital to analog converter (DAC) converts the digital signal back into an analog signal which is then amplified and used to drive the loudspeaker as before.
A Deeper Look into Digital Audio
Just as in computers, the samples in digital audio are stored in ‘words’ which have a fixed number of bits. Therefore digital audio signals are characterized by the rate at which the source is sampled, called the sampling frequency, and the number of bits used to store each sample, called the word length. For example on a CD the original sound was sampled at 44.1kHz (44,100 times per second) and each sample is stored in a word 16 bits long.
Because of two key factors the digital signal is not an ‘exact’ copy of the original source signal:
Sampling: The original signal has a value at any point in time, that is, it is continuous. Since the digital signal is produced from a sequence of samples taken at instants in time, we only know its value at those instants. We don’t know what its value is between the sampling instants.
Quantisation: Quantisation means using a limited set of values to represent something that is continuous.
As the figure shows, the continuous grey signal is being represented by the red digital signal which is limited to only one of 8 values. As a result the values of the samples are only approximately equal to the original signal.
Because a digital signal is just a sequence of numbers it’s easy to use mathematics to manipulate it in many useful ways. This is one of the main attractions of digital signals. Two bits of math allow us to overcome the deficiencies imposed by sampling and quantisation.
Nyquist Sampling Theorem: The Nysquist sampling theorem basically states that once we have sampled the original signal fast enough we can use the samples to create an exact copy of the original signal.
Interpolation: Interpolation uses the samples to estimating the value of the original signal at any time between sampling instants.
Therefore once we sample fast enough and use interpolation we can come very close indeed to reproducing the original signal from the digital audio signal.
© Wayne Butcher