In Reply to: RE: peak level at each octave band posted by dave789 on December 28, 2008 at 12:24:52:
Paradoxical as it sounds, you may need to reduce the input level by more than 4 db. For example, if the input signal were a square wave and you were to boost the fundamental by 4 db and leave all of the overtones unchanged, the peak signal would end up being boosted about 5 db. (Verified by using audio editor software to generate test signal, perform EQ and examine peaks.)
Another way of looking at this would be to see what would happen if you were to leave the fundamental the same and reduce all of the overtones in the input square wave. At first glance you might think that the net result would have lower peaks, but actually, they would be higher. The reason this happens is that the original square wave was a combination of a louder fundamental that has been partially cancelled by the overtones. When you filter out the harmonics you end up seeing the louder fundamental. In the case of a square wave this amounts to a factor of 4 / PI , which is around 2 db. In your 4 db example, the harmonics are reduced, but not eliminated. The magnitude of this anomalous effect will depend on the input waveform. If the input wave form is a pure sine wave then there will be no anomalous effect and a 4 db boost in the selected band will result in a 4 db boost in the output.
While the example was constructed with an artifical waveform, this was just for simplicity. I first encountered this effect with a computer audio workstation when I got a clipping warning when I reduced the amplitude of some frequency bands on a previously unclipped waveform.
It is best to keep a few db of headroom to avoid these problems.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
This post is made possible by the generous support of people like you and our sponsors:
Follow Ups
- It's not quite so simple... - Tony Lauck 11:54:46 01/08/09 (0)