It is a destructive compression method.
Destructive compression is compression that is achieved by losing information. This means that if you unzip the compressed signal using this technique, you will not find the start signal.
It is possible because it exploits the properties of the human ear.
The frequency spectrum of human ear is in the range of 20 Hertz to 20 kilohertz. If a sound contains frequencies outside this range, we can eliminate them without loss of audio quality since the ear does not hear them.
In reality, the human ear is not equally sensitive to all frequencies in the theoretical audio spectrum. its maximum sensitivity is between 2 kilohertz and 5 kilohertz. Outside of this range, you must increase the power of the source to have the same sound sensation.
In order to reduce the size of the recording files, one may decide to eliminate all high frequencies beyond 15 kilohertz.
The mp3 technique uses what is called the masked frequency technique which consists of not recording sounds that have a low sound level compared to the others.
The MP3 uses the Huffman algorithm (1952) at the end of the processing.
The Huffman algorithm is used in all compression algorithms (text file compression, image compression, compression sounds). It is based on the use of a variable length code and the probability of an event (in this case, a frequency) occurring. The more often a frequency appears, the shorter its code (low number of bits to represent it). The file is read for the first time and a table is drawn of the frequencies that appear and the number of times they appear. We deduce the appropriate code. This encoding is used last. This is the final phase of compression. It is a non-destructive encoding.
In total, MP3 first uses the properties of the ear to reduce the size of a song, then it processes the stereo aspect (if the two channels of a stereo sound have similar sequences, instead of duplicating two sequences, we store only one, since the information for the right channel and the left channel are identical) and we end up with Huffman encoding.
The use of all the information reduction possibilities depends on the place you want to give 1 minute from your tablet and, therefore, on the compression rate to be applied.
For encoding MP3 sound files, there is more talk in terms of bit rate than compression rate.
Bit rate is the number of bits authorized in 1 second: the more you want to compress a song (so it takes up as little space as possible), the more the bit rate has to be low to choose.
P more compression and less audio quality is good.
A compressed signal with a bit rate of 64 kbit / s has a bandwidth of 11 kHz (all frequencies beyond 11 kilohertz are suppressed).
To obtain this compression ratio, in addition to suppressing the frequencies, the high frequencies were also transformed into a mono signal instead of keeping them in stereo. The MP3 sound of files compressed with a bit rate of 64 kbits / s has nothing to do with the original (we only use it to listen to portable music players).
