frequency adjustments down, signal up

[jackstpaulUHS]

I use CEP 2.1a, if that matters.

Sometimes when I reduce/cut the volume of a frequency/range--no other changes, the overall signal increases. I don't get it. It would seem to me that reducing a frequency would lower the overall signal to some degree depending on the relative levels to begin and the reductions.

Example: I have a track that has low-level rumble--just noise--120 of below that makes it sound muddy. When I use a scientific filter--great for getting at narrow ranges--and cut the level of that range out, the over all signal goes up. I have the song hard-limited to -.5 and the change kicks a lot of it way over.

I'd think of the overall level as consisting of levels at various frequencies added together: A + B + C +......and that reducing or cutting "C" would lower the overall signal.

I sometimes get a similar problem if I use other EQ types--graphic (I have 30 bands), parametric...

Brought the problem on myself when I didn't know what I was doing and boosted a too-wide freq range that included that low end. The song is just an acoustic and electric guitar and one vocal--no bass, no drums--nothing that registers at that low freq, other than noise.

Sometimes when I've applied other effects--click/pop eliminator, maybe even noise reduction--I've seen increases in level. If a signal is there, how can taking something out boost? I’m not--of my own doing--increasing the level of something else.

I’m not increasing the signal output level as can be done within each type of effect application—I leave it at zero.

Thanks.

 

[SouthSIDE Glen]

One real possibility is that you have some kind of heavy duty phase cancellation going on between two or more tracks in certain frequency ranges. Reduce some of the uncomplimentary phase signal from one or more tracks via EQing or bandpass filtering, and the overall mix level will rise.

G.

 

[Daisy]

I'd think of the overall level as consisting of levels at various frequencies added together: A + B + C +......and that reducing or cutting "C" would lower the overall signal.

If these filters were scientifically perfect, and operated with surgical precision in a completely perfect and symetrical fashion ... then in theory, they would behave the way you're expecting them to.

But thankfully, they don't work that way.

 

[mixsit]

It's real and I've read the explantion -damned if I remenber though.
There is the hump you get at the rotation point on steep filters, maybe it's there to a lesser degree on all of them.

 

[SouthSIDE Glen]

It's real and I've read the explantion -damned if I remenber though.
There is the hump you get at the rotation point on steep filters, maybe it's there to a lesser degree on all of them.

I'm familiar with this on low/high pass filters, I don't know if I've seen it on parametric or notch filters though.

Below is an example: Both plots show the same high pass filter (plug-in) set to the same knee frequency of 85Hz. The top plot has a Q setting of 0.5. The bottom a Q setting of 1.25. Note the "bump" that mixsit is talking about actually is boosting the signal by about 3dB at 100Hz.

G.

 

[mixsit]

Here you go. It's more than I want to tackle right about now.
http://recforums.prosoundweb.com/index.php/m/186814/0/?srch=filter+cut+peaks#msg_186814

 

[jackstpaulUHS]

Thanks.

Basically what I get out of all this is that it's a normal occurrence and nothing wrong/off with what I've done/my system.

That the hardcore details of it are matters of electrical engineering, which I have no background in.

That the best way to think about it is to not think about it all knowing what I know now, but to put it in terms related to my initial mathematical model it's not like this:

The overall signal db is a not a simple function of A+B+C+D.....

Rather the freq variables aren't independent, they interact with each other so that changing (observed result)C doesn't necessarily affect only C, rather it can affect (observed result) A and/or/not B and/or/not D.....

Hence whatever values result from f(A,B) f (B,C) f(A,C) f(A,B,C) f(A,B,D).....and on produce the final outcome from whatever mathematical relationships they inhere.

Obviously there exist upper and lower bounds--maxima and minima--for any given set of variables and attendant functions in terms of the resulting outcomes, but that's not important to me because all I need to care about at my level of need-to-know is that, yes, as you here have helped me understand, what I've observed is natural. When I witness it in the future, don't worry about it.

This is the most extended glimpse I’ve had about the elec. engineering, the mathematics and physics, involved, and I’m quite impressed by the sophistication and intelligence of many of you. I taught myself how to record, apply effects, mix, basic mastering without any clue about the more complicated nature of this stuff—and what a lot of the shorthand terms stand for, even if familiar with the phenomena they identify. It makes me wonder how much I’ve been missing that would help even in a not very complicated way to get better outcomes.

I’ve never come across a simple-enough while detailed-enough tutorial on the issues in general, having been left with FAQs, Wiki, and manuals for various components that often aren’t very worthwhile.

Are there any good tutorial/manuals/sites to be found that fit the bill?

An audio engineering Wiki of its own would be a great idea rather than bouncing around within the regular Wiki system.

 

[NYMorningstar]

Bob Katz has some good books.

 

[SouthSIDE Glen]

but that's not important to me because all I need to care about at my level of need-to-know is that, yes, as you here have helped me understand, what I've observed is natural. When I witness it in the future, don't worry about it.

A very dangerous position to take.

I'm not saying that you have to me Mr. Electrical Engineer and understand all the details. Frankly, 90% of that stuff in that engineering thread is probably not even applicable to your situation, or if it is, is on a level of detail that you don't need to know.

But frankly, what you observed is something who's possible primary causes can be indicative of other problems. In the case of possible phase masking, it certainly would raise some red flags in my head as to why that phase masking was there to begin with and whether it should be there at all. In the case of a low-pass or high-pass filter bump, sure it is natural for the filter to work that way, but that doesn't mean you should just accept it. There are more ways to skin that bandpass cat that don't require accepting the bump.

You don't need a PhD in physics to do this stuff well, and I would never push some of that crap on anybody. But in this instance I think you might be setting the "need to know" bar a bit low by simply accepting it and moving on, because you might be missing something important.

I’ve never come across a simple-enough while detailed-enough tutorial on the issues in general, having been left with FAQs, Wiki, and manuals for various components that often aren’t very worthwhile.

Are there any good tutorial/manuals/sites to be found that fit the bill?

An audio engineering Wiki of its own would be a great idea rather than bouncing around within the regular Wiki system.

Check out the address under my name for a few tools and an increasing number of articles you may find helpful. Also included on that site is an interactive book catalog with over 90 titles related to recording and audio engineering and links to full detail description pages for each of them.

As NY'Star said, Bob Katz has a good book for plunging into the engineering detail called "Mastering Audio". If that's too technical or too much to chew, I think that "Understanding Audio" by Daniel Thompson strikes a very good balance.

Also, check out a site called The Recording Project for a pretty large archive of free articles of various quality.

G.

 

[RAMI]

Is it not possible that you have the mix compressed, so when you reduce something, the compression doesn't work as hard, thus allowing the signal to rise in volume....I have no idea if that makes any sense at all.

 

[SouthSIDE Glen]

Is it not possible that you have the mix compressed, so when you reduce something, the compression doesn't work as hard, thus allowing the signal to rise in volume....I have no idea if that makes any sense at all.

Well, first of all, the compression would have to be real-time and not already applied to the mix; second, it would have to be chained after the EQ.

But even then - unless one has some kind of weird side-chaining thing set up maybe - the compressor is still going to be set up for X amount of gain reduction above a threshold of Y, and neither of those values are changed by any changes made to the signal before compression. By your theory the compressor is not working "less hard", it's just dealing with potentially smaller peaks. Compressing smaller peaks will not result in larger peaks after compression, but even smaller peaks yet.

G.

 

[mixsit]

Is it not possible that you have the mix compressed, so when you reduce something, the compression doesn't work as hard, thus allowing the signal to rise in volume....I have no idea if that makes any sense at all.

I could see a possible way -maybe. If the comp was tilted to be sensitive in one area (boosted to the detector) and that's where you cut.

 

[RAMI]

By your theory

You're giving me way too much credit, my friend. That wasn't a theory...it was a shot in the dark.

 

[SouthSIDE Glen]

You're giving me way too much credit, my friend. That wasn't a theory...it was a shot in the dark.

Hey, man, that's how most theories start out .

G.