All About Digital Audio - http://www.soundonsound.com
An analogue audio signal is so called because some physical property is used to store or convey a representation of the original air pressure changes associated with a sound wave.
For example, the grooves of a vinyl record or the electrical voltages within audio equipment or produced by a tape recorder reflect precisely the changing nature of the original sound.
In other words, the record groove or the electrical signal is an analogue of the variations in air pressure.
An analogue signal is continuous – it changes from moment to moment, but does so smoothly and without breaks or gaps. Furthermore, its content is not innately constrained in any way.
It can change at any rate it likes (theoretically, it can have an infinitely wide frequency response) and have any size (infinite dynamic range).
The practicalities of earthbound physics and human hearing mechanisms mean that we generally restrict the frequency and dynamic ranges to something that can be conveniently accommodated by the recording medium, transmission format or the equipment itself, but it is still worth remembering that there are no absolute limits (or behavioural rules) inherent in an analogue signal.
This very lack of rules is the main disadvantage of analogue signals, because it makes them very susceptible to degradation of various kinds.
The two most obvious examples are noise and distortion, both of which change the signal waveform in a way which our ears can detect very easily, but which remain perfectly allowable under the 'rules'.
Indeed, some sounds that we would consider musical involve noise (cymbals and snare drums, for example) and distortion (such as that produced by overdriven guitar amps), and are clearly wanted effects – not degradation at all!
Unfortunately, although our ears can spot such degradations all too easily, it would be virtually impossible to build an analogue machine which could automatically detect and remove unwanted noise and distortion from an analogue signal, because it would not have a human appreciation of what is intended (ie. subjectively pleasant to human ears) and what is unwanted.
The same applies to wow and flutter – a perennial problem in analogue systems.
How can a machine tell the difference between intended vibrato effects and unwanted wow or flutter?
Probably the greatest advantage of digital audio signals is that they are bounded by certain well defined rules about their size, shape, and how fast they can change.
They are also discontinuous, because in digital recording, an original analogue waveform is measured at specific time intervals and its amplitude at each of these points stored.
This results in a string of numbers which depicts the waveform in its development over time, rather than representing it by the continually changing property in an analogue recording medium.
It may seem odd that the digital representation of an analogue signal is discontinuous, and you might think that something must be missing. Well it is; all the frequencies above 20kHz, in the case of a CD recording.
More on this in a moment.
Because digital signals are constrained by well defined rules, degradations which break those rules (such as waveform distortion, added noise, or timing variations) can be detected and removed without altering the audio content of the signal.
Although there are various alternative systems (some of which we will examine in later articles), the vast majority of digital audio systems encode the numbers which represent the original audio waveform as binary data, by means of a process known as pulse code modulation (PCM), which we will examine in more detail next month.
It might seem deliberately obscure to record these all-important numbers in a different counting system to decimal, the one we're all used to, but there is a reason: in binary counting, there are only two values, zero and one, and all numbers are therefore represented by strings of 0s and 1s.
Binary is ideally suited to our available technology, because these two states (0 and 1) can be represented as the on or off status of a transistor switch, north or south magnetisation on magnetic tapes, bumps or flats on CDs, high or low voltages in electronic circuits, or light and darkness on optical media.
A square wave is used in PCM encoding; a point halfway up the wave is defined such that signals below are classed as 0s and signals above as 1s.
Even if the PCM signal is distorted or noisy, it can always be regenerated (as long as the degradation is not too severe), so that the data representing the actual audio waveform can be recovered undamaged.
This, of course preserves the quality of the encoded audio exactly as it was when it was first sampled.
This fundamental characteristic of digital systems means that the quality of a digital audio signal is not primarily dependent on the physical medium used to store or transport it, but on the initial conversion from the analogue domain to the digital one, and the subsequent reconstruction to an analogue form for replay.
The most important part of a digital audio system is the analogue-to-digital (A-D) converter, because it is here that the audio quality is defined and fixed. If information is lost at this stage, nothing can be done to recover it later.
One digital-to-analogue (D-A) converter may make a better job of recreating the analogue signal from the digital data stream than another, but both are limited to working with the data as supplied by the initial A-D conversion – which is why this first stage is so important.
1ª Parte - http://www.soundonsound.com/sos/may98/articles/digital.html
2ª Parte - http://www.soundonsound.com/sos/jun98/articles/digital2.html
3ª Parte - http://www.soundonsound.com/sos/jul98/articles/digitalbasics3.html
4ª Parte - http://www.soundonsound.com/sos/aug98/articles/digitalbasics.html
5ª Parte - http://www.soundonsound.com/sos/sep98/articles/digitalbasics.html
6ª Parte - http://www.soundonsound.com/sos/oct98/articles/digbasics.html