Warning Reason Filters

10 Replies, 4924 Views

I just noticed that the low pass filter (both built into the nn-xt and the standalone one) noticeably boost frequencies over 15k. since it is on by default in the nn-xt i've caused this havok on almost all of my tracks Icon_sad

to solve it i tried the sound forge filter instead and it mutes the highs a bit at the default setting. at least with the soundforge there is a wet/dry which avoids the pop you get with switching reason's from bypass to on Roll

ok onto the some news. im going to re-record an audio sample thru my philips sound card until it sounds wrecked Lol
Sounds like possibly frequency warping as it approaches Nyquist to me.

To my understanding, a filter/eq's response has to be unity at Nyquist (it's the r00lz Teef ) and this fucks the frequency response up, especially as you get up top.

To put it another way;

Quote:Frequency-warping problems are common when using the bilinear transform to map continuous-time filters onto the z-plane. The bilinear transform is convenient to use, because it preserves filter order and filter stability. With the bilinear transform, there is no aliasing of continuous-time frequency, but the continuous frequency axis is warped as it is being mapped onto the discrete-time frequency axis. Warping gets progressively worse at higher frequencies.

The simplest method for reduction of bilinear-induced warping is to use oversampling. Oversampling adds expense to algorithms, because the discrete-time filters must run at higher rates. Additionally, the resampling filters tend to be computationally expensive. However, oversampling minimizes frequency warping by placing the range of human hearing in a lower, and relatively less-warped, portion of the discrete-time spectrum. Oversampling is convenient, because the resampling filters can be designed independently of the filters to be discretized.

From here;

http://www.uaudio.com/webzine/2008/septe...nalog.html

There's probably more about... when you see the curves and the difference oversampling makes, it is clear as fook.
[Image: protabl3.gif]
Don Cherry Wrote:Every human is blessed in her or his life with one love (passion), no matter how long it may last. This Absolute love will last in one's heart and soul forever.
Not noticed anything untoward in any of my tunes to my ears, bar the filters just being completely weak and not warm in any way shape or form.

Then again I work in a completely untechnical way Grin
Macc Wrote:Sounds like possibly frequency warping as it approaches Nyquist to me.

To my understanding, a filter/eq's response has to be unity at Nyquist (it's the r00lz Teef ) and this fucks the frequency response up, especially as you get up top.

To put it another way;

Quote:Frequency-warping problems are common when using the bilinear transform to map continuous-time filters onto the z-plane. The bilinear transform is convenient to use, because it preserves filter order and filter stability. With the bilinear transform, there is no aliasing of continuous-time frequency, but the continuous frequency axis is warped as it is being mapped onto the discrete-time frequency axis. Warping gets progressively worse at higher frequencies.

The simplest method for reduction of bilinear-induced warping is to use oversampling. Oversampling adds expense to algorithms, because the discrete-time filters must run at higher rates. Additionally, the resampling filters tend to be computationally expensive. However, oversampling minimizes frequency warping by placing the range of human hearing in a lower, and relatively less-warped, portion of the discrete-time spectrum. Oversampling is convenient, because the resampling filters can be designed independently of the filters to be discretized.

From here;

http://www.uaudio.com/webzine/2008/septe...nalog.html
Ooh, they talk about eigenfunctions. Go eigenfunctions!

This is quite good if you like that sort of thing:
http://en.wikipedia.org/wiki/Bilinear_transform
right, im familiar with filters, i just assumed that if it was on by default and the filter is set to 22k, they might just bypass the damn thing :/
and thanks for the link. bookmarked it to read on my lunch break Smile
Slothrop Wrote:Go eigenfunctions!

Lol
djfada Wrote:
Slothrop Wrote:Go eigenfunctions!

Lol
Seriously, my life at the moment is modular eigenfunctions of the Laplacian on the hyperbolic upper half plane. Ie it's all this sort of stuff that Macc talks about with fourier series except that I'm working over the complex numbers on the surface of some sort of weird quotient-torus and it has no conceivable application to making a better sounding EQ.

It's getting *seriously* *fucking* *boring*.
I remember that the time they first mentioned eigenfunctions was almost exactly I gave up on doing well in my degree.

Sounds like a German electro band.
[Image: protabl3.gif]
Don Cherry Wrote:Every human is blessed in her or his life with one love (passion), no matter how long it may last. This Absolute love will last in one's heart and soul forever.
I suddenly realized the other day why it is that for a cycle of a waveform you can talk in terms of its Fourier series but for a generic signal you have to talk in terms of its Fourier transform and it's basically down to a theorem about the Laplacian having pure point spectrum on a compact set ie it's all about eigenfunctions ie Lovesmilie
IE WTF
[Image: protabl3.gif]
Don Cherry Wrote:Every human is blessed in her or his life with one love (passion), no matter how long it may last. This Absolute love will last in one's heart and soul forever.

Possibly Related Threads…
Thread Author Replies Views Last Post
  Reason Rack Extensions denaris 3 4,395 17th July 2012, 09:29
Last Post: denaris
  Frequency analyser reason 4 brandz 5 3,610 22nd February 2011, 15:23
Last Post: brandz
  Making beats in Reason 3.0 safetyboy 16 6,779 5th September 2007, 14:03
Last Post: cordani
  Reason talk Macc 52 13,744 1st April 2007, 10:56
Last Post: subvert
  working with reason siLLy KiD 23 6,681 26th September 2006, 22:56
Last Post: socialengineer