Digital sound encoding
The development of methods for encoding audio information as well as moving images (animation and video recordings) occurred with a delay relative to the types of information discussed above.
A computer is a digital device, that is, an electronic device in which a discrete signal is the operating signal. Today’s computers operate on discrete signals that carry binary values, conventionally designated as “yes” and “no” (at the electrical level: 0 volts and V volts, for some non-zero value of V). With a one-step binary signal, you can transfer information about one of two positions: 0 (“yes”) or 1 (“no”). Using N binary signals in one step, you can transfer information about one of 2 N positions (2 N is the number of combinations of zeros and ones for N signals). The interaction of all the blocks that make up a computer occurs through the exchange and processing of one or more binary signals simultaneously. They are all control codes as well as the information that is processed itself, everything is represented on the computer in the form of numbers. For this reason, audio signals in digital equipment are also represented as numbers.
So how can you describe an analog audio signal in digital form? A real audio signal is a complex waveform, a certain complex dependence of the amplitude of a sound wave in time. In Fig. 2 shows a graph of a real sound wave.
For computer processing, an analog signal must somehow be converted to a sequence of binary numbers. Let’s proceed as follows. We will measure the voltage at regular intervals and write the obtained values into the computer memory. This process is called sampling (or digitization).
Converting an analog audio signal to digital is called analog-to-digital conversion or digitizing. The process of this transformation consists of:
carry out measurements of the amplitude of an analog signal with a certain time interval: sampling,
subsequent recording of the amplitude values obtained in numerical form – quantification.
The time sampling process is the process of obtaining the instantaneous values of an analog signal converted into a specific time step, called a sampling step.
The higher the sample rate (that is, the number of samples per second) and the more digits assigned to each sample, the more accurately the sound will be represented. But this also increases the size of the sound file. Therefore, depending on the nature of the sound, the requirements for its quality and the amount of memory occupied, some compromise values are chosen.
The number of signal measurements taken in one second is called the sample rate or sample rate, or sample rate (from English “sampling”). Obviously, the smaller the sampling step, the higher the sampling frequency (that is, more often amplitude values) and therefore the more accurate representation of the signal we get.
The human ear does not notice the gradation of the received signal. Here the following analogy can be drawn. Each person watched movies in the cinema and before their eyes on the screen there was a continuous and fluid action: but, in fact, a filmstrip is a series of still and discrete images that move at a high speed of 24 frames per second . Since human eyes are characterized by a certain inertia, they are easy to fool, which the filmmakers use extremely cleverly. Our ears are also somewhat imperfect and can be tricked in this way, representing a continuous analog signal as a sequence of rapidly changing instantaneous voltage values. But unlike a film strip, changing the “sound frame” happens thousands of times faster.
Now, to record each individual amplitude value, it must be rounded to the nearest quantization level. This process is called amplitude quantization. In more formal terms, amplitude quantization is the process of replacing the actual (measured) values of the signal’s amplitude with values that approximate with some precision. Each of the 2 N possible levels is called the quantization level, and the distance between the two closest quantization levels is called the quantization step. Quantization of signal values introduces additional interference into the signal spectrum, called quantization noise or division noise … Quantization noise (error) refers to the signal that makes the difference between the signals reconstructed original and digital audio tracks. This difference results from the rounding of the measured signal values.
