Notch vs. band in EQ work

Dec 10, 2012
1,427
11
38
Washington, DC
I'm mainly thinking about this in the context of the eternal kick vs. bass battle, but are there major downsides, in phase weirdness or otherwise, to using a notch on an instrument to completely carve out a frequency, as opposed to just using a tight bell curve with a large amount of db reduction?

I've experimented with creating a narrow-ish notch wherever the kick drum peaks and removing those frequencies entirely from the bass guitar to make room for the kick, but I was wondering if that's a bad idea and I should just be doing db reduction with a normal bell instead. Thoughts?
 
In this particular case I'd rather use a normal band, since a very narrow cut in your bass guitar is unlikely to provide enough space for your kick to breathe. Of course it depends on the source... but I don't use notches that often since the amount of reduction cannot be controlled, and I use them more like a "quick and dirty" fix (e.g. to kill a fizzy peak or two on distorted guitars or basses, especially impulse-based ones) than as an actual clean and non-destructive solution...
 
I've never used a notch filter while mixing. It's a problem solving tool, to use for instance when you get cicada noises doing field recording work or getting some kind of localized, resonant hum out of your tracks. Not much utility for it during mixing.
 
I also use notches very rarely. mathematically speaking notch filter is a band-filter with gain set to minus infinity dB. In reality numbers in computers aren't precise enough to achieve that perfectly - especially in low frequencies. You get -80dB at most if your EQ uses max-precision available in modern processors (so with most plugins usually around -40dB). So only difference between a notch and -40dB cut would be the curve shape (depending on the actual algorithm/circuit used by your EQ).
I also agree with everything stated above. Band filter gives you a little more freedom with minimal sacrifice (actually none if your EQ supports cuts down to -40dB).

A little math. Cut of -24dB decreases amplitude to 0.061= cca. 6%. If you add that to original 106% is roughly 0.5dB increase. So cutting more than 20dB is basically pointless because you'll not hear a difference in the mix anyway (except maybe more phasing issues).