Acoustic sagas. How the sound was turned into figures?
Shortcomings of analog carriers of a sound together with the relentless course of progress led the person to thought that the sound can be presented in a type of information for record on digital carriers. The same fate did not pass any object which is giving in to the quantitative and qualitative description in system of visual and sound parameters today. What general principles of transformation of an analog sound into sequence of figures?
The analog signal by means of ATsP (it is analog - a digitizer) breaks into small sites lasting several hundred-thousandth fractions of a second. On each of such sites of ATsP measurement of amplitude of a sound is performed. Size, the return to length of these sites, is called the frequency of sampling and is measured, as well as sound frequency, in hertz. Obviously, than it is higher (than more often we measure amplitude), especially we will be able to save full information on a signal.
Too frequent measurements, however, demand a lot of place for storage of the obtained information. On the other hand, reduction of frequency of sampling leads to loss of information which is contained in a signal. How to choose the optimum frequency of sampling? The answer is given by the famous theorem of Kotelnikov according to which the sound with a frequency f cannot be digitized with a sampling frequency below than 2f . It means, for example, that attempt of digitization of a sound frequency, say, of 1000 Hz with a frequency of sampling of 1000 Hz is doomed to failure. The silence, or the distorted sound having nothing in common with initial will be result. Establishment of a sufficient top limit of value of frequency of sampling is an important consequence from this theorem.
The standard upper bound of frequency of the sound perceived by our ear is equal to 20 000 Hz. Therefore, the top sufficient limit of frequency of sampling of a sound can be taken for 40 000 Hz. In reality sampling frequency at a sound recording on CD makes 44 100 Hz. The professional studio equipment is calculated on work with a sound with a frequency of sampling of 48 000 Hz. The standard of frequency of sampling for phone line is taken for 8 000 Hz. It is quite enough to save information which is contained in a voice message and also to keep emotional coloring and to provide recognition of a voice.
On it history with sampling does not terminate. Breaking a sound signal into sites of a certain length and measuring amplitude of fluctuations on these sites, you should not forget what can also matter this amplitude at record in memory of the computer only strictly certain discrete values. It means that the quantity of steps between the maximum and minimum values of amplitude is limited. Their height (a difference between the next values of amplitude) depends on quantity of these steps and, therefore, the their is more, the quality of a sound is higher.
Features of storage of information in memory of the computer led 256, 65536, 16777216 and 4 294 967 296 pieces to quantity of possible steps. Correspond to these figures 8 - 16 - 24 - and 32 - bit sounds (the corresponding degrees of number two as the computer works with a binary code). Of course, increase “ bitnost “ a sound leads to increase in amount of information. However 16 - usually it is quite enough bit sound (65536 options of amplitude of a sound) for high-quality record of the majority of sound signals.
Now we with ease will be able to count how many places will be taken by a three-minute stereosignal at record on usual musical a compact - a disk. We multiply 180 seconds by 44 100 Hz, then by 2 bytes (it corresponds to 16 bits), then for 2 (the number of channels, as a signal stereophonic). Total figure - 31 752 000 bytes, or about 30,3 MB of information. It is a lot of, the truth? Whether it is possible to reduce without noticeable loss of quality? It is possible. But it is a subject absolutely of other article.]