When music began to digitize, going from analog to digital, problems were introduced that had existed since the 70s.
Using a 16-bit depth bit introduces noise into the recording. There are people who say that it does not make a difference to digitize to 16 bits or 24 bits and do the wrong calculation. Actually, there is a difference and it is fundamentally the noise it generates.
On the other hand, it was discovered a long time ago that listening to the loudest music in terms of volume gives the impression that there is an improvement in quality.
Hence, in the 70s, what is today known as “the volume war” was born.
Although it is completely true that there is a big difference in that some music sounds more or less loud in volume and that it is necessary to normalize it.
Also what is if the normalization is high gamma (high quality) as it can only be done by the Mp4Gain, which normalizes using the RMS per band (each sound band is normalized separately using the RMS -Root Mean Squared-), this achieves an improvement in sound of more than 35% percent.
Normalizing by peaks is a thing of the past, even RMS normalization is outdated unless a banding is introduced and maskings are removed and disguised, in addition to a harmonic reconstruct.
When encoding music, what is sought is to eliminate information and a good part of this information is that produced in masking, where a louder sound covers the weaker sounds and these are eliminated when encoding, but this supposes a loss of information that will result in a loss of quality.
Harmonics are also eliminated (sounds that represent mathematical multiples of the main sound that sound very low, but help to give a special timbre to each sound).
Then the modern normalize like the one that gets MP4 Gain which is done by RMS and by band and with reconstruction of harmonics and trying to reverse the masking, produces a noticeable improvement in quality.
