I wanted to ask you if you know and use iZotope RX, specifically the Waveform Statistics function, which I use to check the True peak level; from what I understand, if it has a positive value, it's not good.. The fact is that most of the tracks I inserted into Waveform return a positive peak value and consequently several clipped samples that escape my ear.. (this even when analyzing famous songs, tracks downloaded from iTunes or taken from CDs).. now, it seems to me that everything can be resolved by normalizing, still using iZotope, to -0.33... but I don't understand why? can you help me?
It is an audio signal interpolation problem at the time of analysis. I realize that to understand it well, you should have studied some signal theory. In any case, the software interpolates the waveform only according to Nyquist, whereas in hardware, interpolation also occurs along all other intermediate points of the wave. Therefore, in the software case, the measurement scale is subject to variations; since the measurement is not very precise, there is a margin of error to account for. The "out of threshold" is probably within the margin of error (I don't think it exceeds +0.5 dB). If you used hardware dedicated to level measurement, it would most likely be within the 0 dBu limits, so perfectly normal. To avoid these problems, for those who do not have a dedicated level measurement device, it is common practice to put a limiter at -0.3 dB on the master. Once, it used to be set at 0.0 or -0.1 dB.
Regarding the "threshold" you are talking about, which should not exceed +0.5 dB (margin of error), in reality most of the tracks I have submitted for analysis return a positive value that goes far beyond that.
I always set the limiter to -0.3 as you say; in my opinion, these waveform statistics are not very reliable, I don't know what other explanation to find!
Is "Signal Theory" always a topic related to sound engineering?
