How much compresses an MP3
MP3 compression was an engineering response to the problem of digital storage and its large memory resource requirements. A conventional digital signal called PCM (Pulse Code Modulation) could easily require up to 10 Megabytes of memory per minute. This would represent about 30 Mb for a three minute song.
That requirement for storage memory could be handled by any computer if it were a few files, but when talking about three thousand songs the numbers become worrying. As if this were not enough, there is the problem of the Internet and its current transmission speeds. In the case of telephone lines, they have a limitation on their transmission bandwidth, so very large or heavy files represent a problem for conventional network traffic.
MPEG3 compression is considered the sound part of the original MPEG1 format that was intended for cinematography. Its abbreviations, Moving Picture Experts Group come from the committee that was created by the ISO Organization (international Standards Organization) and IEC ((International Electrotechnical Commission) to develop this format. Its principle is based on the Psychoacoustic model.
The human ear is known to discriminate sound according to its limitations. According to subject matter expert Paul Sellars, “If you hear solitary applause in a room, it will surely sound loud, but if it is preceded by the sound of a gunshot, it will sound fainter. The same thing happens in a room when you record a rock band, at a certain moment the strongest sound guitar in the mix, until the moment the drummer plays a certain cymbal, at which point the guitar will seem to attenuate “This phenomenon is used by the MP3 algorithm to perform its compression . I once explained it in the article that talked about ATRAC compression of the Minidisc.
The MP3 format divides the sound into 32 sub-bands, which allows it, according to the Psychoacoustic model on which it is based, to give priority to one element over another. At a certain moment in the material we can have a predominant low frequency sound of the kick drum, a high frequency of the cymbal and the vocalist at the same time. The algorithm is not that it eliminates two of them, but that it dedicates less storage space to them.
The mathematical part used with MP3 compression goes through the Shannon-Nyquist theorem, which states that for a wave to be properly reproduced in PCM digital format, its frequency of takes (Sampléo) must be twice the highest that is want to reproduce. In this case if we want to reproduce the frequency of 22.5KHz, (The auditory range oscillates between 20Hz-20KHz), our sampling frequency should be 44.1KHz.
The Fast Fourier Transform (FFT) is also used, which as we know can decompose a complex wave (PCM material) into a fundamental wave with its harmonics, all from its amplitude. The Discrete Cosine Transform is also used, which is based on the FFT but only using the real numbers
UNTIL IT IS RECOMMENDED
These formats will continue to be perfected and emerge, but it should be understood that despite being disseminated there may be details that will not be perceived. In other words, for serious Audio work this format should not be used.
Some improvements can be made by looking for compressors that have a better ratio, such as 224, 256 and 320 Kbps. You can also consider using VBR (Variable Bit Rate) encoding where musical passages with greater dynamic complexity are treated with a higher rate. storage in contrast to the simplest. However, this will bring other complications because not all the reproducers can handle them.
