![\"NORMALIZE](\"http:\/\/voyard.free.fr\/textes_audio\/mots_du_son\/images\/signal_original.gif\")
<\/p>\n<\/p>\n
The “Normalize” feature has the function of scanning all or part of a selected audio file and detecting the highest peak.<\/p>\n
In fact, it passes all samples in the file one after the other (counting at a rate of 44, 48 or 96000 per second of recording), temporarily stores the value until it finds a sample, the highest value ignores all the lowest values \u200b\u200bthat the last.<\/p>\n
Once this final value is detected, you can imagine a value of -8 dB, the software applies an increase of 8 dB to all the samples in the file and therefore raises the total level of the file by 8 dB to find the main peaks at zero dB.<\/p>\n
![\"Normalize](\"http:\/\/voyard.free.fr\/textes_audio\/mots_du_son\/images\/signal_normalise_-6_RMS.gif\")
<\/p>\n
Signal normalized to 0 dB peak or 0 dB at Absolute Peak – Listen<\/p>\n
The advantage? Normalization does not allow “partial” to systematically change the listening volume from one song to another or from one fragment to another, from one disc to another. This feature says what it does, normalizes (bring it to the default).<\/p>\n
Note: In general, a maximum normalization level of zero dB can be defined. For example: -1 dB. To prevent some intermediate CD player converters or some output amplifiers from these players from being distorted (?). Subject to debate.<\/p>\n
Advantage: This processing only changes the overall level of the file, but by no means its dynamics (it does not reduce the difference between the highest and lowest levels of recording). So nothing to do with a limiter, a compressor or an expander.<\/p>\n
Problem: if only one of the samples in the recording or the selected part reaches 0 dB (either positive or negative), even if all the others are -20 dB or less, the system will consider that the file does not need to be changed and then normalization will still be strictly ineffective, which is kind of silly, right?<\/p>\n
Original signal with 0dB peak – Listen<\/p>\n
Therefore, before normalizing the recording or marking, make sure that some isolated peaks do not deceive the system and, if necessary, remove or reduce them with the “volume” function or another (select the peak, reduce or redraw with the pencil tool) .<\/p>\n
Also, remember to select your file in parts to apply appropriate “volume” or “normalization” levels for the sounds to be normalized.<\/p>\n
Danger 2 (Return): It is good when the software allows it not to normalize certain recordings or fragments of recordings to zero dB.<\/p>\n
For example: Record steps on the lawn whose average level is -35 dB and less, which is perfectly correct. If you “normalize” it to ZERO, you end up with an outrageous sound that will sound like anything but footsteps on the grass.<\/p>\n
Mistrust: Some software offers various options for “normalization”: either in “RMS” or “Peak”. The two processes must be clearly differentiated.<\/p>\n
RMS: It has nothing more to do with it. In fact, the software will apply a compression algorithm to the processed file. For SF, it will have 3 other predefined values, but you can choose any other value.<\/p>\n
Peak: The software performs a “normalization” according to the technique described above. Therefore, there is no change in the recording dynamics.<\/p>\n","protected":false},"excerpt":{"rendered":"