Unexpected Filter Problems

D

DannyGantastic

Guest
Using Fruity's built in parametric EQ, I've been having trouble attempting to remove headroom-grabbing sub-bass frequencies.

When I add lowpass, notch, or -12 dB peaking filters at 10hz, and set the resonance below the point where any ampification occurs at the cutoff frequency (Determined graphically inside the EQ plugin), but before higher frequencies are diminished much, the following happens:

-My sub-bass frequencies diminish somewhat , and nothing is is particularly affected (According to my spectrum analyzer)
-All my low end to around 150 hz sounds a lot weaker
-The overall sound, however, is about 3dB louder than before filtering is applied.

My first question is: Are these results contradictory? What's going on here?

I can understand that Fruity's filters aren't going to be of the best quality, and my spectrum analyzer isn't spectacular either, but how can a sound be a great deal quieter to my ears, almost identical on my spectrum analyzer, yet be fully 3dB louder than the unfiltered sound?

My second question is: How would you reccommend I go about removing frequencies under 20 Hz without substantially distorting the sound above that threshold?

Thansk for anyone's input.
 
When I add lowpass, notch, or -12 dB peaking filters
Click to expand...

You need to use a high pass filter set at around 20-40hz with no resonance (and I'd suggest a 24db/octave filter roll-off).

This will cut significantly reduce all frequencies below the cutoff frequency and allow all higher frequencies to pass, hence the name.

Tom
 
Sorry, when i said low-pass I meant high-pass. I had decided three hours prior to the post that I needed to sleep, and I guess that typo is proof :)

Anyway, that's precisely what I'm doing -- The problem is that I'm getting the ill effects both of having too much resonance and having too little, at the same time; Bass frequencies up to around 150 are being seriously attenuated with areas, and the sub-bass below the cutoff frequency is diminished somewhat, but at the same time, the overal signal is 3 dB louder than without filtering.

Thanks, and sorry again for the typo.

Any ideas?
 
Ok that makes sense, I thought it was a bit of a strange question :-)

How about inserting a compressor (preferably multiband) post-filter to control the dynamics better? You could use moderate compression to even out the level. Are you treating a whole mix or a single element/submix?
 
It's just a single monophonic bassline that I'm working with. And in reference to the filter I'm using: FL doesn't state how many poles it has.

Rather than find a fix to the problem, I'm more curious as to learning (for future reference, an general understanding) why it is that it sounds quieter, the spectrum analyzer shows virtually no difference, yet the sound has 3dB less headroom.

Unfortunately, the only multiband compressor in my arsenal is a freebee which can't divide between bands under 87 Hz. The main problem with compressing normally has been that since the output already sounds quieter, that compressing to bring the peak amplitudes to their original level makes the quietness even more severe. I'll keep trying :)

Thanks for the advice so far!
 
If theres no resonance then I have absolutely no idea how a filter is reducing headroom !?!

Non-resonant filters are purely subtractive. I'm as confused as you I'm afraid!

Its not something silly like the spectrum analyser has only coarse calibration and so isn't displaying small amplitude changes in each frequency band clearly?

Maybe the filter itself has an input/output volume that needs to be turned up? I'm not familiar with FL but I know some filters have an input drive to achieve distortion, perhaps its turned down too much, if you have one?
 
Well, at least it's good to know I'm not missing something too obvious :)

I doubt it's a problem of the analyzer having too few bands -- It has more than I can count:
http://www.elementalaudio.com/Images/insnoalrm.jpg

And I doubt it has anything to do with input boosting -- The filter I've been using is one built into FL's parametric EQ, not likely to be intended for distortion.

I'm still confuzzled -- But thanks for all your assistance! The time you've spent scratching your head over my little conundrum is much appreciated =).
 
No problem, I'm actually sat on my arse with a broken ankle at the moment - so scratching my head is one of the things I can do . . . my studio is in an attic room up two flights of stairs errgggghhh

Good luck anyway,

Tom
 
I've experienced many situations in which high-pass filtering can actually make the signal louder (as, I guess, the peaks and troughs of the low-frequency waves are subtracted, allowing higher-frequency transients to reflect their true level.)

So this isn't unusual.

Have you considered high-pass-filtering each track individually, rather than the entire mix? This'll give you less surprises when it comes time to mix the tracks.

-Hoax
 
The audio stream in question is just a monophonic bassline.

Could you please elaborate more on your suspicion as to why the signal is louder? Thanks.
 
Well, it's not so much a suspicion or theory as it is an observed phenomenon.

I did extensive audio editing within Alchemy (an early Mac sound editor), and I found that often as I'd add a highpass filter at 10 or 20 Hz, the signal would get significantly louder. I generally had to process the file with a negative gain before applying the EQ.

Later, I moved to Peak as an audio editor, and found the same thing. In fact, when previewing Waves EQs before applying them, I found a sub-bass frequency cut to bring up the overall level quite a lot of the time. I generally had to reduce the input of the EQ.

The "low-frequency peaks cancel out higher-frequency peaks, level-wise" theory seems to be pretty weak, though it's what occurred to me first when thinking about this.

But it seems to be generally true, through different software, and different filter/EQ algorithms.

Must be some facet of physics I haven't understood as of yet.

Still true, though.

-Hoax
 
Last edited by a moderator:
Alright, thanks for the tip! Answers the bulk of my question to hear that others have had the same problem.

And letting me know that it's been true over a wide range of filters saves me from trying tons of different filters until I get something that didn't decrease headroom.

Thanks, psychotope and Cruel Hoax, for your input. A burning question has been (mostly) satisfied :)
 
You must log in or register to reply here.
Back
Top