Question on snare peaks

Just wondering if anyone would be able to explain why this is happening.

I have only really noticed this with snares. I will have a snare hitting at say -10db, I noticed after putting a eq on the snare rolling of the lows gently my snare will be hitting at say -8?? I would have thought if anything cutting out frequencies would make the snare hit at a lower volume?

Cutting the frequencies doesn’t make sounds quieter like that. For any given frequency there is a corresponding wave in air and that wave has an amplitude which in some convoluted way corresponds to the dB reading. Reducing the amplitude of a particular frequency will reduce the volume. So if you have two different frequencies playing at the same volume, cutting one out entirely won’t change the overall volume.
As for your situation, define soft? It could be that the way your rolling things out results in ‘spikes’ that you can’t see on the screen but are affecting the sound. For instance, an LP filter with a low bandwidth adds volume at the cutoff frequency. High width will reduce volume.
Dunno what you use but this could still apply:
FL parametric EQ 2 (and probably 1 too) is pretty straightforward in that you can just ‘reset’ the width knob, by right clicking, to produce a ‘flat’ (ie no spikes) cutoff.

TL:dr
Your EQ is probably adding gain at the frequencies you’re rolling at.

Are you using a highpass? The problem as Jesslem pointed out is probably the resonance. Boosting the resonance of a highpass at the fundamental (around 200Hz for a snare) may improve you sound though.

Like a logarithmic function?

Not true?

A Q of 0.7071 (or sine of 45 degrees) gives a neutral reponse at the cutoff. A butterworth filter also has this response.

no jess got it mixed up

it can get louder when you remove the low rumble because it was hiding the peak slightly

it means some hi frequency suddenly had a shorter way to travel

thats related to the understanding of transient peaks

the thing OP, is that digital meters cant be trusted … actually to be fully straight up . . the read out on the master HAS to be trusted but the individual track meter was a leftover from real studios and doesnt actually tell you its red when its red - and similarly the other way around… like it could be red lining but not show up

so just read the masters read out

2 Likes

what you are seeing is phase rotation from the group delay by using a filter of a specific number of orders increases the group delay, a good way to think about eq is a delay working on a fractional level and the group delay as feedback,if you think about it that way,the peaks make sense and you’ll never hear this reflected as a change in magnitude at a given frequency,though you will perceive it as such when put up against another sound (because the phase has been rotated and the timing of the frequencies altered),it doesn’t boost the magnitude,it alters the phase hence why it doesn’t always result in a peak,it usually will when there is allot of low end information that was once there as low frequencies take more cycles to produce their energy.you increase the problem quite abit if you use filters with high orders (steep curves) hence why phase integrity goes completely out the window in most cases with filters that use FFT with allot of windowing

2 Likes

good post but not entirely related

we’re talking about an individual sound that without context becomes louder with the addition of an eq

so its definitely about the transient

Yes a high pass filter and yeah around 200hz. So your saying if I use a filter with a Q of 0.7071? There will be no resonant peak around the 200hz??? I’m not at home at the minute but I did try this with a narrow Q and a sharp Q and was still hitting louder then when there was no eq at all.

He never said anything about him being able to perceive the difference in volume,just that he can see a 2db increase on a peak meter.my explanation would explain that.you could simply confirm by looking in an oscilloscope at the phase relationships before and after.i’m willing to bet that the frequencies where the transient is located will be VISUALLY be an order of magnitude greater than it was before he added the eq

Thanks, I got a little confused technically with this response. I am talking about a snare on its own though. Can you please explain what you mean group delay as the feedback?

:badteeth:

Tbh m8 its just not what matters in this case. i dont want to step on your tech talk.

BUT

Theres very little chance that this ISNT a case of playing a sound on a speaker that cant represent or go low enough to monitor the rumble

rumble is by definition surplus junk - so its also the bit of amplitude a program would have a hard time measuring correctly - - theres also the unknown stray freqs down there that are just mud

but most progs deal with transients and read those peaks instead

theres even inaudible low rumble that the graphic read outs wont pick up on. btw

1 BigUp

you should read this again

it can get louder when you remove the low rumble because it was hiding a bit that was on its way to peak slightly

removing the surplus rumble - means some of the frequencies contained in the sound suddenly had a shorter way to travel - which meant the meter managed to pick up on it - more clearly

thats related to the understanding of transient peaks

the thing OP, is that digital meters cant be trusted … actually to be fully straight up . . the read out on the master HAS to be trusted but the individual track meter was a leftover from real studios and doesnt actually tell you its red when its red - and similarly the other way around… like it could be red lining but not show up

so just read the masters read out and dont worry too much with a sound by sound basis

1 BigUp

Okay, thanks for the response.

1 BigUp

To summarise/clarify, I think the suggestion was that if it isn’t resonance nor inter-sample peaks, it may be the manipulation of the phase relationships between sinusoidal waves (within the same sound, not in context) of the Fourier analysis which are introduced by the filter that are now summing in the same direction, increasing the overall amplitude.

See the sections on Fourier analysis and filters.
http://www.sp4comm.org/webversion.html

1 BigUp

Fair, I guess the kid who explained it to me way back when was about as confused as I’ve been since, haha.

On the frequency thing, that was more or less something I figured myself, but how’s that not true? It’s basic interference isn’t it?

bless everyone. Thanks.

At some point the waves will sum to be at or close to the sum of their absolute amplitudes.

Normal EQs cause phasing at the cutoff point. Try using linear-phase eq and you notice that the volume doesn’t jump anymore (not much at least, it might still go from -10dB to let’s say -9.5 or something). You get annoying pre-ringing however.

it’s because you don’t get my “techtalk” that you are not understanding my answer.do the same thing with a linear phase filter and tell me if the same thing happens (the answer is it doesn’t)

I’d be all for your argument if the OP said he could HEAR the change but he said he can only SEE it in a peak meter.therefore it’s some of the frequencies being shifted out of their original alignment and possibly slightly rubbing up against one another

You are talking about inter-sample peaks here (an entirely different thing).this only really applies to anything around about 0dB full scale,so in no way is it relevant to what the OP is seeing