million dollar scientific EQ ? [Thread from 2007 bumped]

[DRDRDR]

Hello guys & gals !

I'm looking for an EQ with the steepest possible curve
and the ability to choose very precisely the HZ limit.

Ex : if i choose to cut exactly below 72 HZ, i'd like no 71 HZ at all in the resulting sound.

look at my little drawing :



any exact recommendation ? thanks !
best regards,
DRDRDR

[laputa_sync]

Perhaps a filter like betabugs bugpass would be better...?

[CinningBao]

check out SpaceBoy's FreEq Boy (not free but darn cheap)
http://www.elevayta.net/product3.htm
it might not have the frequency resolution you desire though..

[ouroboros]

yes, it looks like you need a cutoff filter.
otherwise an eq with a shelf and at least 24db slope.

[tony tony chopper]

note that an EQ with the steepest slope does NOT sound good

[Cyrosis]

http://www.kvraudio.com/get/2671.html

ReaFIR should do what your looking for (and then some).

http://img228.imageshack.us/my.php?image=refiroq0.jpg

[jackson]

Nice picture

Stack a bunch of filters with the same settings on them, I'm pretty sure the way it works is that each one will make the q steeper. At least I think that's what I heard.

[Kingston]

I haven't actually seen any proper linearphase 100dB super drops available as plugins yet. I've been playing with them in matlab, using as steep slope a 64bit math gives in. Fun toys and weird sounding as it will simply KILL everything below/above a certain point.

Can't see much use for one outside research. I seem to recall they had a gaussian one (also from matlab) in some early nebula beta as the volterra algorithm precision tester.

[laputa_sync]

isnt the idea to achieve a balance in the time and frequency domains>?

[ToneCarver]

Voxengo Redunoise (with noise reduction disabled) might get you pretty close:

http://www.kvraudio.com/forum/viewtopic.php?t=173508

[kylen]

Electri-Q has steep 82db/oct slopes or something like that...

FreEQ Boy lets you import a text file that can say something like 72Hz == 0db, 71 Hz == -120db. It's a steep thing - I forget exactly the format but I've used it and it is a steep one 96db/oct or something to that affect.

For the million dollar EQ with 130/db slopes or whatever, you might need a cement lined 1Km circular FFT buried under your house (or better yet the guys house next door, I think they're radioactive ).

[laputa_sync]

FFT is probably the better choice over IIR in terms of slope steepness.

[kylen]

I was making steep crossovers in a routable host for multi-band processing when I used FreEQ - the 3 FreEQs nulled perfectly.

[dalor]

blocc hiPass and blocc loPass - both with 192 db/Octave(!) is all you ever need.

I wouldn't use an FFT based EQ, they often introduce artefacts.

Just my 2C: the way you'd like to cut off frequencies would make any sound most unnatural. I'm not sure if this analogy makes sense, but when doing picture manipultation or collages where you insert elements of another picture into your composition, you'd leave a feathered/antialiased edge as well to make elements blend in together, even if its 1 pixel. This 1 pixel could be referred to the dB/Octave in the audio world. You'd need to leave some feathering into other areas to blend in with the rest.

[Christian Budde]

To have an infinite steep filter you'd need infinite time! Or in other word's it is not possible. You always have to truncate and with this you'll face the uncertainty relation. Either you'd have a good frequency resolution (e.g steep cutoff) or you'll have a good time resolution (think of transients). All FFT based filters may have a perfect curve exactly *AT* the bins (which are displayed), but usually you can't say what's in between (ripple). If you use a window function you could smooth the frequency response by the expense of ripple in the time domain.
It's the same trade-off with IIR filters. Either you use an elliptic or chebyshev filter (which means ripple in the frequency response), or something like bessel/butterworth (less steep->you'll need a higher order->ripple in time domain).
This is the scientifical point of view. The musical point of view is, that it sounds bad. It's really sounds too artificial. It's because your ear normally doesn't expect such sharp filters (there are no discontinuities in the nature!)

Some ideas,

Christian

Btw.: Electi-Q also has the Bugpass integrated and offers up to 108 dB/Oct Butterworth filters. It would also be possible to supply any other (higher) cutoff, but it really doesn't make sense.
But if you'd really want to offer a million dollar...