Subject:Paragraphic EQ
Posted by: HenryM
Date:6/22/1999 1:36:00 PM
I use the Paragraphic EQ in Sound Forge to remove background noise at 60 Hz and 120 Hz (both at 0.3 octave, -25dB). I also use a low shelf at 80 Hz (to -inf. dB). The filter is great for removing the noise, but if the file has already been normalized, I must reduce the wet output gain, some times by 2.2 dB. (I don't use dry output at all.) I am only attenuating with the paragraphic EQ. Why is the gain being boosted? Any insights or suggestions are most welcome. (I realize that if I filter before normalizing, the problem is reduced, but I don't understand comceptually why the Paragraphic EQ would behave in this way.) |
Subject:Re: Paragraphic EQ
Reply by: Sonic
Date:7/2/1999 12:08:00 PM
Hi Mark, In order to get reasonable rolloff on the shelves at minus infinity gain, we used a not-quite-textbook approach to the EQs and specifically the shelves. There is a little bit of peaking near the cutoff of the low shelf (high pass) filter. The notch filters actually make up for this a little, but if this is in a low audible region you can experience a couple dB of gain. As a side note, the Track EQ that ships with Vegas takes a different approach. It doesn't exhibit this behavior, but it's stopband rejection isn't quite as good. If you ever decide to get Vegas, this EQ will work in other DirectX apps like Sound Forge (as long as Vegas is installed). We are always looking to make our products perform better. Thanks for the input. Jason. Mark Henrichsen wrote: >>I use the Paragraphic EQ in Sound Forge to remove >>background noise at 60 Hz and 120 Hz (both at 0.3 octave, >>-25dB). I also use a low shelf at 80 Hz (to -inf. dB). >> >>The filter is great for removing the noise, but if the file >>has already been normalized, I must reduce the wet output >>gain, some times by 2.2 dB. (I don't use dry output at all.) >>I am only attenuating with the paragraphic EQ. >> >>Why is the gain being boosted? >> >>Any insights or suggestions are most welcome. >> >>(I realize that if I filter before normalizing, the problem >>is reduced, but I don't understand comceptually why the >>Paragraphic EQ would behave in this way.) |