MP3 from the inside: psychophysiology of sound
This format uses an extremely complex encoding algorithm. Unlike conventional filing cabinets, which need to compress information in such a way that after file extraction not a single bit is changed, MP3 has slightly different goals.
In addition to mathematical compression algorithms, this format also contains a very complex algorithm for removing unnecessary audio information, based on the psychological and physiological characteristics of the human body. I will try to stop at this point in more detail.
As said, MP3 is a streaming format. This means that the audio information during encoding is divided into equal parts of length, which are called frames. All frames are mutually independent. Each of these frames is encoded separately with its own parameters and has a header in which these parameters are described. When played back, a sequence of frames decoded and produces a continuous sound from the recorded audio.
What are the benefits of this approach? First of all, the ability to rewind, as you can easily jump to an arbitrary frame and play the sound from this point. Second, it is this structural feature that makes MP3 a truly networked format. Once the first frames are loaded into RAM memory or disk cache, the player starts playing them, while simultaneously loading new frames, thus achieving playback continuity. And finally, if you couldn’t download the entire MP3 file from the internet, then that’s okay, you can still listen to music, only the player will get to the point where the connection was cut off and stop.
So let’s get back to our frames. With high quality MP3, and this is ~ 320 kbs bit rate, only mathematical compression algorithms are used to encode frames. At the same time, the quality does not suffer at all, but the size is reduced only four times, that is, the compression ratio is the same as that of a normal filing cabinet; that is why MP3 files are practically not compressed by normal file cabinets. When the bandwidth (bitrate) is reduced to 256 kbs and less, the same algorithms come into play to eliminate “unnecessary” sounds that are based on the characteristics of the perception of sound by the human ear, the so-called “psychoacoustic model”. “Unnecessary” elimination processes quantification. The lower the bit rate, the more difficult the quantization will be.
What are the criteria for evaluating the “necessity” and “uselessness” of sounds? The vast majority of codecs emit sounds that are considered outside the threshold of human hearing. In this case, the threshold value, so to speak from fakto, is taken as a value equal to 16 kHz. Despite the fact that this threshold is recognized as an elementary value and is included in all physics textbooks, this approach is incorrect. People are very diverse in their physiological characteristics. In addition, it must be taken into account that the hearing threshold of young people is much higher than that of older people and can easily exceed this average value. Much also depends on the strength of the signal. Therefore,
Another criterion by which the “futility” of sound is evaluated is a condition based on a characteristic of human hearing such as the inability of most people to distinguish signals that are below a certain power level, and this level is different for different frequency ranges. When using the psychoacoustic MP3 coding model, CODEC automatically outputs inaudible low power frequencies. Unfortunately, again, people are not the same and those who are able to accurately distinguish these frequencies often complain of loss of sound quality during encoding, while the average majority does not notice it.
But the most important feature of the MP3 psychoacoustic coding model is the so-called masking effect. It is thanks to this effect that it is possible to compress both the original audio data. The essence of this effect is that a weak signal from one frequency range is often masked by a stronger signal from an adjacent range, if present in an audio recording, or by a strong signal from the previous frame. This strong signal causes the ear to temporarily desensitize the current frame signal. In fact, the phenomenon of “temporary stunning” occurs. For each sound range, the amount of masking effect, generated by the signal from adjacent ranges and the signal from the previous frame, is determined.
