This is called the Gibbs phenomenon, and it happens whenever you use an idealized low pass (brickwall) filter. Basically, whenever you use a brickwall filter, the removal of higher frequencies causes the resulting signal to be discontinuous at small scales (meaning it suddenly jumps in some areas) and that adds strange artifacts to it (hence the ringing)
The filter is self-resonating. The steeper your curve, the more the frequencies at the bend of your curve will be affected using traditional filters. If you use a high pass filter on the same kick sample this will be easier to recognize and visualize.
filters can cause weird phase shift issues sometimes too so set them by ear because even a filter set at 20 hz will effect stuff above that frequency sound wise.
The steeper the curve, the greater the potential issues such as artifacts or problems with the resulting resonance. The brickwall EQ isn't completely useless, but it's more of a "peace of mind" thing, I guess. Don't use it to cut into strong, audible sound or you'll get that strong ringing -- use it in areas that appear dead to ensure they are dead. (Or don't, since we were getting on fine without brickwall EQ already!) A lot of producers would frown on your 48Q curves too, because those also ring a bit, but if the results sound okay to you then go for it.
Thought i'd explain whats actually going on (did signal processing in my engineering degree): essentially, if you created a plain sine wave at say 100Hz, and kept adding odd harmonics you end up with a square wave right? by ADDING frequencies you create an on-off type of signal to one that previously was a nice smooth forward and back motion. Now think about if you had a nice LOW signal (e.g. 5hz) which was off most of the time and turned volume ON when it was at it's peak for the same duration as the 100Hz squarewave 'on' cycle, you'd end up with a short sample that was very similar with a slight curvature but essentially the same. Then you took that short sample and looped it at 100hz.. Which means turning something on/off MUST have added additional harmonics throughout. You can say the opposite is true also, the higher harmonics are what is causing the on/off of the square wave function. The same is true for this kick sample, the original waveform of the kick has alot of harmonic content that 'turns it on/off' at the start of the sample, and the end of its tail, by deleting this content willy-nilly you alter all of the volume automation that creates the original envelope, thus 'forcing' it to propagate and also have a delayed onset.
Also to add, a true brickwall filter will have a negative infinity value at x in front of the filter, so the harmonic content is essentially deleted. The 48Q filter, while being sharp, will still let 'some' content through, and so has less of an effect. you might think 'well its very little really, because look how sharp it is' but you still let ALL frequencies through past, so from 58Hz to 20kHz if all frequencies are playing, new harmonics are 'added' by the very fact that you are filtering with a steep curve, you end up with some ability for the sample to still volume automate as you would expect and the tail is not as prominent.
ALSO again to add, if you check out what oldmate has selected at 3:44 and hit 'delete' and thought you had created a brickwall with no issues, you'd be wrong. By deleting the tail, you have reintroduced higher harmonics, soz mates.
What's the difference between these two scenarios? First scenario you have a sine wave, then you add one harmonic, then you turn off the added harmonic so you're back to a pure sine wave. Second scenario you EQ out the harmonic with a brickwall EQ curve.
First scenario: unless your 'pure' sine wave is permanently 'on' forever, it will have harmonics when it starts/finishes. Adding and subtracting different harmonics will affect how it starts and finishes necessarily. Second scenario: EQing with a brickwall will take out not just your added harmonic, but also the other harmonics that turn the original sound on/off (volume automation) so you'll get your ringing. The best approach to getting rid of 'specific' harmonics that you dont want is to not generate them in the first place (use additive synthesis i.e. Harmor or similar). There will STILL be other harmonics though due to the volume envelope
hmm I wonder how steep the slope on the "filter" in Additive Drums happens to be. I guess it doesn't matter if the end result sounds great. Sometimes people want a ringing snare, for instance. Linear phase can create ringing too but I don't know how serious a "problem" it always is.
Just use a normal high pass filter as steep as 48db/octave...don't use brickwall (or 98db/octave if you're using fab-filter Pro-Q2)..basically, nothing too unnaturally steep.
When making purely digital sounds a too loud base might create interesting harmonics. I mute the lower bands on a compressor to remove the base because it sounds cleaner. Am I right?
What if I'm not hearing what I think I'm supposed to hear...? I don't hear anything coming out of the program. Is it supposed to be like this? Recording problem? *_EDIT:_* For anyone not hearing anything, put on headphones (not earphones, headphones) or get some speakers with bass active. Or you will not hear anything like I wasn't with my laptop's speakers hahaha.
Love your tutorials, channel, overall cool personality and sense of humor. thanks for all your time and knowledge.. is there any chance you can help me learn how to make one of the following songs 1. dynamo city - one night in hackney 2.stefano noferini - this is fucking house music 3. ben sims - re manipulator (adam beyer remix) thanks =)
RoboticusMusic cause your ears can only hear to 30hz to 25hz or something after that you cant hear it anymore. So most of the time people cut the low frequencies to 30hz if the eq lets them.
I get the feeling that's a secondary thought and it has something to do with how EQ works, or the algorithms and delays added. For example if you want to boost 5hz sharply you're out of luck with almost all EQ's, besides that this is usually not a typical thing to do with an EQ. It also makes you wonder what happens at either end of the EQ below 30hz or above 20khz when you boost around those areas. What happens beyond what the chart is showing you? Do these EQ's have an internal point, say at 35khz that you can't see on the graph where the EQ simply stops responding to a high boost that slopes past the 20khz end of the EQ graph?
That's not the problem. By default, the highest frequencies in a hit occur at the impact point of the sound. That's what makes a drum a drum and not a tone, and those freqs are all well above 58Hz.
Awesome Video as always! Its just an interesting though which you dont have to use but the whole concept/idea of this gives me another point of view. thanks !
Just use it for the right purpose. For instance, if you've got an area of high frequencies that you definitely don't need, and the area is seemingly dead but you can't tell if you're just not able to hear it track by track, that's what the brickwall EQ is for. It's intended to cut out the guesswork, so that you don't have to think, "Hey, I wonder if that seemingly empty highs area is actually stacking up into a problem when ten sounds are playing at once?". Only use it in areas of sound that are seemingly "dead" - don't use it to sculpt area with high amplitude material. Remember that the problem of ringing that you hear in this example is an extreme version of what's happening in all such EQ curves, and the hidden resonance is just worse with steeper and steeper curves. Treat ALL steep curves with suspicion and listen carefully to what they're doing, and use them sparingly.
Is this guy trying to difame izotope or he just doesn't know that he is using the brickwall filter in "analog mode"? You don't need to use another plugin, just switch to "digital mode" and you won't get those artifacts.
Guys WTF..................... This is about phase problem, NOTHING ELSE! Try Linear Phase and all of this bad stuff will dissapear.i Hope that helps :)
This is called the Gibbs phenomenon, and it happens whenever you use an idealized low pass (brickwall) filter. Basically, whenever you use a brickwall filter, the removal of higher frequencies causes the resulting signal to be discontinuous at small scales (meaning it suddenly jumps in some areas) and that adds strange artifacts to it (hence the ringing)
The filter is self-resonating. The steeper your curve, the more the frequencies at the bend of your curve will be affected using traditional filters. If you use a high pass filter on the same kick sample this will be easier to recognize and visualize.
filters can cause weird phase shift issues sometimes too so set them by ear because even a filter set at 20 hz will effect stuff above that frequency sound wise.
The steeper the curve, the greater the potential issues such as artifacts or problems with the resulting resonance. The brickwall EQ isn't completely useless, but it's more of a "peace of mind" thing, I guess. Don't use it to cut into strong, audible sound or you'll get that strong ringing -- use it in areas that appear dead to ensure they are dead. (Or don't, since we were getting on fine without brickwall EQ already!) A lot of producers would frown on your 48Q curves too, because those also ring a bit, but if the results sound okay to you then go for it.
Thought i'd explain whats actually going on (did signal processing in my engineering degree): essentially, if you created a plain sine wave at say 100Hz, and kept adding odd harmonics you end up with a square wave right? by ADDING frequencies you create an on-off type of signal to one that previously was a nice smooth forward and back motion. Now think about if you had a nice LOW signal (e.g. 5hz) which was off most of the time and turned volume ON when it was at it's peak for the same duration as the 100Hz squarewave 'on' cycle, you'd end up with a short sample that was very similar with a slight curvature but essentially the same. Then you took that short sample and looped it at 100hz.. Which means turning something on/off MUST have added additional harmonics throughout. You can say the opposite is true also, the higher harmonics are what is causing the on/off of the square wave function. The same is true for this kick sample, the original waveform of the kick has alot of harmonic content that 'turns it on/off' at the start of the sample, and the end of its tail, by deleting this content willy-nilly you alter all of the volume automation that creates the original envelope, thus 'forcing' it to propagate and also have a delayed onset.
Also to add, a true brickwall filter will have a negative infinity value at x in front of the filter, so the harmonic content is essentially deleted. The 48Q filter, while being sharp, will still let 'some' content through, and so has less of an effect. you might think 'well its very little really, because look how sharp it is' but you still let ALL frequencies through past, so from 58Hz to 20kHz if all frequencies are playing, new harmonics are 'added' by the very fact that you are filtering with a steep curve, you end up with some ability for the sample to still volume automate as you would expect and the tail is not as prominent.
ALSO again to add, if you check out what oldmate has selected at 3:44 and hit 'delete' and thought you had created a brickwall with no issues, you'd be wrong. By deleting the tail, you have reintroduced higher harmonics, soz mates.
What's the difference between these two scenarios?
First scenario you have a sine wave, then you add one harmonic, then you turn off the added harmonic so you're back to a pure sine wave.
Second scenario you EQ out the harmonic with a brickwall EQ curve.
First scenario: unless your 'pure' sine wave is permanently 'on' forever, it will have harmonics when it starts/finishes. Adding and subtracting different harmonics will affect how it starts and finishes necessarily.
Second scenario: EQing with a brickwall will take out not just your added harmonic, but also the other harmonics that turn the original sound on/off (volume automation) so you'll get your ringing.
The best approach to getting rid of 'specific' harmonics that you dont want is to not generate them in the first place (use additive synthesis i.e. Harmor or similar). There will STILL be other harmonics though due to the volume envelope
Thanks Multiplier for the explanation!
hmm I wonder how steep the slope on the "filter" in Additive Drums happens to be. I guess it doesn't matter if the end result sounds great. Sometimes people want a ringing snare, for instance. Linear phase can create ringing too but I don't know how serious a "problem" it always is.
@Multiplier What about in mastering when you cut the very lows and the very highs does it have negative effects in that case...?
I was thinking the same, I usually cut with a brick wall below 30hz.
+Kaisui Man, Isn't 30Hz a bit too high? 20Hz is a bit safter.
Naaa, if you use a special EQ for that call "Linear Phase".
Just use a normal high pass filter as steep as 48db/octave...don't use brickwall (or 98db/octave if you're using fab-filter Pro-Q2)..basically, nothing too unnaturally steep.
And as carlos said...use linear phase eq
it's just resonance post-ringing
then how is it useful?
When making purely digital sounds a too loud base might create interesting harmonics. I mute the lower bands on a compressor to remove the base because it sounds cleaner. Am I right?
Any nonlinearity causes distortion of some kind. Otherwise it's making a compromise somewhere to achieve linear phase
Multiplier, could you do this test with TDR NOVA? IIRC they have an uncommon kind of filter behind the HPF and LPF.
72dB/oct is very steep
But the steepest you can get aside from brickwall is 120dB/oct
Have you done this experiment with a Linear Phase EQ?
What if I'm not hearing what I think I'm supposed to hear...? I don't hear anything coming out of the program. Is it supposed to be like this? Recording problem?
*_EDIT:_* For anyone not hearing anything, put on headphones (not earphones, headphones) or get some speakers with bass active. Or you will not hear anything like I wasn't with my laptop's speakers hahaha.
Love your tutorials, channel, overall cool personality and sense of humor. thanks for all your time and knowledge.. is there any chance you can help me learn how to make one of the following songs
1. dynamo city - one night in hackney
2.stefano noferini - this is fucking house music
3. ben sims - re manipulator (adam beyer remix)
thanks =)
Why does my brain instantly want to do the opposite of what you say. Just to be different
Probably because of "Wowsers trousers"
issadopesounda
is it possible to do a super steep filter with rea fir on gate mode?
Why don't EQ's typically go below 30hz?
RoboticusMusic cause your ears can only hear to 30hz to 25hz or something after that you cant hear it anymore. So most of the time people cut the low frequencies to 30hz if the eq lets them.
I get the feeling that's a secondary thought and it has something to do with how EQ works, or the algorithms and delays added. For example if you want to boost 5hz sharply you're out of luck with almost all EQ's, besides that this is usually not a typical thing to do with an EQ. It also makes you wonder what happens at either end of the EQ below 30hz or above 20khz when you boost around those areas. What happens beyond what the chart is showing you? Do these EQ's have an internal point, say at 35khz that you can't see on the graph where the EQ simply stops responding to a high boost that slopes past the 20khz end of the EQ graph?
Thats interesting. Thanks! :D
Also attack is lost completely
That's not the problem. By default, the highest frequencies in a hit occur at the impact point of the sound. That's what makes a drum a drum and not a tone, and those freqs are all well above 58Hz.
The attack is the fight frequency click that the filtered out with the 58Hz lowpass.
Awesome Video as always! Its just an interesting though which you dont have to use but the whole concept/idea of this gives me another point of view. thanks !
That’s mad
Isn't that a plugin bug tho? Does this happen on Engineers Filter as well?
I get that added tail when using Analogue Brickwall but when using Digital Brickwall it looks fine, what's the difference?
The artifacts are introduced in the adc process. Digital filters are free to operate with perfect accuracy on a sample by sample basis.
i've literally used the brickwall eq in every one of my tracks haha i guess i'll stop now
Just use it for the right purpose. For instance, if you've got an area of high frequencies that you definitely don't need, and the area is seemingly dead but you can't tell if you're just not able to hear it track by track, that's what the brickwall EQ is for. It's intended to cut out the guesswork, so that you don't have to think, "Hey, I wonder if that seemingly empty highs area is actually stacking up into a problem when ten sounds are playing at once?". Only use it in areas of sound that are seemingly "dead" - don't use it to sculpt area with high amplitude material. Remember that the problem of ringing that you hear in this example is an extreme version of what's happening in all such EQ curves, and the hidden resonance is just worse with steeper and steeper curves. Treat ALL steep curves with suspicion and listen carefully to what they're doing, and use them sparingly.
brickwall lowpass the master at 16kHz ONLY. make sure its on digital mode or else it undoes limiting then you have to damage your audio twice as much.
Is this guy trying to difame izotope or he just doesn't know that he is using the brickwall filter in "analog mode"? You don't need to use another plugin, just switch to "digital mode" and you won't get those artifacts.
*defame
Guys WTF..................... This is about phase problem, NOTHING ELSE! Try Linear Phase and all of this bad stuff will dissapear.i Hope that helps :)
Nope it's about why you shouldn't use infinitely steep filter slope.