Filter Question

[CableChannel]

When I put on MFilter, Q defaults to 0.60 for a 12db filter. Yet when I right click on it, it jumps to 0.71. So what is the right default value here?

The reason I ask is: if I have a highpass and want it to be inactive, I set freq to 20kHz and Q to -> 0.60 or 0.71? Would either of these values actually cause a little boost just below the cutoff freq?

[CableChannel]

Unfortunately the Equalizer tutorial of Melda does not answer this. But I just read something on Wiki:

0.5 is the aperiodic damping limit of an oscillation which reaches equilibrium fastest without overshoot. So that should be the ideal Q factor for an "inactive" filter that should be inaudible (e.g. highpass at 20kHz), right?

[MeldaProduction]

Well, MFilter initially opens an active preset, which probably uses Q=0.6. There's no good answer here.
Technically 0.71 is 1/sqrt(2), which is the value where a 2 pole low-pass filter reaches -3dB at the cutoff frequency. Well I suppose that doesn't really help does it? Anyway this value is more or less the limit point before the filter starts being resonant.
I don't understand the reason though - you said HP at 20kHz, well, that would more or less filter out everything. And what does it even mean "inactive"? Why don't you just disable the band?

[CableChannel]

Well, MFilter initially opens an active preset, which probably uses Q=0.6. There's no good answer here.
Technically 0.71 is 1/sqrt(2), which is the value where a 2 pole low-pass filter reaches -3dB at the cutoff frequency. Well I suppose that doesn't really help does it? Anyway this value is more or less the limit point before the filter starts being resonant.
I don't understand the reason though - you said HP at 20kHz, well, that would more or less filter out everything. And what does it even mean "inactive"? Why don't you just disable the band?

Sorry, me dumb, I meant lowpass

I put a lowpass into a track which should do a little sweep down/up in the middle of the track. The rest of the track it should do nothing to the audio.

So I wondered if it is ok to leave a lowpass filter active with cutoff = 20kHz. If I choose Q too low it will audibly damp frequencies below 20kHz, if I choose Q too high it will boost frequencies just below 20kHz. So I really have to choose the right Q so that the filter does not change anything below 20kHz. Or is that actually impossible?

Would recommend automating the bypass button to really disable the filter?

[MeldaProduction]

Well, it actually is impossible. While there's a filter, it will always be doing something. But to be honest, if you use 0.71 Q with 24dB/oct LP and place it at 20kHz, nobody will hear anything (unless you tell them it's there, then their brains may start to "feel" something ).

Actually I was recently checking response of my ears (plus some other people) and found out it's actually much worse than I thought . Basically I personally hardly hear sines above 17k, in presence of othersignals it's even worse, and these signals are usually attenuated for a reason - if they are audible, they just sound horrible . I was always surprised how some people were talking about the "sparkle". Just try a steep high-shelf and amplify everything above 10k by say +20dB, it's sooo painful .
And other finding was that dynamic resolution of 40dB is actually hell of a lot. Try a rectangular tremolo on a noise, or 400Hz sine or just anything... and let it switch between 0dB and -40dB. Or even better, take 2 signals, close enough, with levels 40dB from each other... it's really hard to hear the silent one...
Ears suck .

[CableChannel]

Well, it actually is impossible. While there's a filter, it will always be doing something. But to be honest, if you use 0.71 Q with 24dB/oct LP and place it at 20kHz, nobody will hear anything (unless you tell them it's there, then their brains may start to "feel" something ).

Yeah I don't hear it either, but yet I am paranoid that it reduces the brightness and that sparkle you mention, the felt (esoteric) clarity in sound Anyway I can just use the wet/dry controller to really turn the filter of when not in use.

Basically I personally hardly hear sines above 17k

Either you are young or you have excellent ears. I don't hear anything above 14k

[MeldaProduction]

You can train it I think. I was testing a simple thing:
White noise -> LP filter 120dB/oct jumping 20kHz <-> x kHz. Idea is if you hear a difference. I ended at about 12k. Then I was listening to this tremolo sines, got about 17k and then I heard the difference in the noises to about 16k . Ears are quite adjustable . Anyway it's just enough to have the "sparkle" at say 16k... above it, most people don't hear anything. And even in these frequencies it needs to be really loud to be heard...

[robertszalapski]

Great discussion! If you're paranoid that the boost is going to do something bad, the mix amount is good to bypass the filter entirely, but Vojtech is right that it's probably not an issue.

I like using this site: http://www.audiocheck.net/blindtests_fr ... php?frq=16
That one is easy for me but hard for many. You can go higher or lower depending on your skill level. They have some other tests on there that are interesting as well.

[MeldaProduction]

Uuuuu in this test I ended up at 14k, 10/10, but 15k no difference . How did you do???

[bbaggins]

How depressing. I had to go down to 12 kHz to get 10/10. I guess I should just start recording at 22.05 KHz and save the disk space.

[CableChannel]

How depressing. I had to go down to 12 kHz to get 10/10. I guess I should just start recording at 22.05 KHz and save the disk space.

[CableChannel]

Great discussion! If you're paranoid that the boost is going to do something bad, the mix amount is good to bypass the filter entirely, but Vojtech is right that it's probably not an issue.

I like using this site: http://www.audiocheck.net/blindtests_fr ... php?frq=16
That one is easy for me but hard for many. You can go higher or lower depending on your skill level. They have some other tests on there that are interesting as well.

That noise test is reeeeeeeeally difficult. At 12k it is really difficult for me. 11k works ok.

With sines, I can hear 14k. Listening to a -3db sine file (stereo volume knob set to my usual music listening volume) is even a little painful. Whereas 15k I can't hear at that volume, I have to turn up my stereo to hear it.

[MeldaProduction]

Yeah, this is called "a painful sparkle"

[robertszalapski]

The noise test is difficult. I used Melda's oscillator and tremolo to create some of my own sound files for testing my hearing, and used foobar ABX comparator for blind ABX testing. I can hear 20k sine tones. 19/20 correct. (I worked in 96kS/s to avoid potential aliasing artifacts.) But on the noise test, I can barely hear 18k, only sometimes getting 9/10 or better. So you can hear sine tones about 3kHz higher in frequency that you can hear the filter in that test.

I'm not quite 30 yet and I actually wore hearing protection when mowing the lawn, but my point is, very few people are going to hear a slight EQ bump at around 20k. I can't. The only thing I'd be concerned about is if the bump at high frequency increases aliasing from non-linear processing down the chain, which could be unpleasant. But even aliasing tends to be fine in most real-world situations... I've ABXed some pretty aggressive real-world style use of stuff like MLimiter and not needed the oversampling... no difference audible to me.

My point being, unless you are wanting an extremely resonant filter sweep, I wouldn't worry about it. If you have a pretty resonant one, then sure, make sure it's not resonant when you don't want to hear it. But no one will be disturbed by a slight rolloff of 20kHz. Air does that.

[CableChannel]

I just looked up that at the age of 35 the average hearing capability is up to 15k. At the age of 60 it is only 5k!!!!

Guess all sound engineers have to be rich by the age of about 50 then? Or be retrained to do a whole new job, say lawnmower