INTERACT FORUM

Please login or register.

Login with username, password and session length
Advanced search  
Pages: [1]   Go Down

Author Topic: Bug - high pass filter (v19.0.108)  (Read 2094 times)

slayerizer

  • Recent member
  • *
  • Posts: 6
Bug - high pass filter (v19.0.108)
« on: January 26, 2014, 11:35:46 am »

I found a bug, it could have been there in previous versions. When I go set-up a high-pass filter of 21000 (slope 6db/octave) it works as expected, I can listen to high-frequencies. If I select a value of 21830hz (or higher), It will disengage and play ALL frequencies. I would have expected it to work until 22050hz, or at least hear no sound at that point.

This only happen on 44100hz material. As long as the range is between 1-21829hz, it works. Any higher value and it will give you the full range.

 ::)

Not really a show stopper but seems like a bug!

One dirty way to get around this is to change the sample rate in the OUTPUT FORMAT. If I upscale 44100 to 88200 (or higher), it works as expected.

Logged

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10935
Re: Bug - high pass filter (v19.0.108)
« Reply #1 on: January 26, 2014, 11:46:11 am »

Interesting finding. I'll try to have a look at it, but it might be a bit.
Logged
~ nevcairiel
~ Author of LAV Filters

slayerizer

  • Recent member
  • *
  • Posts: 6
Re: Bug - high pass filter (v19.0.108)
« Reply #2 on: January 26, 2014, 11:55:24 am »

Interesting finding. I'll try to have a look at it, but it might be a bit.

for debugging purpose, I also tested a high-pass filter of 43660hz and get the same behavior on 88200hz content. A filter of 43659hz is working as expected.
Even if I don't have a use for such high-frequencies, it's interesting to see that there is a pattern behind the bug.

I can get around by resampling, just wanted you guys to be aware of it.
Logged

mwillems

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 5234
  • "Linux Merit Badge" Recipient
Re: Bug - high pass filter (v19.0.108)
« Reply #3 on: January 26, 2014, 12:13:57 pm »

for debugging purpose, I also tested a high-pass filter of 43660hz and get the same behavior on 88200hz content. A filter of 43659hz is working as expected.
Even if I don't have a use for such high-frequencies, it's interesting to see that there is a pattern behind the bug.

I can get around by resampling, just wanted you guys to be aware of it.

It probably doesn't have enough "space" to create the correct filter shape; the roll off begins before the setpoint (the filter is already -3dB at the set frequency), so it's probably just failing when it doesn't have enough bandwidth to work with.  Does using a steeper filter slope affect the "failure point"?  I expect it would.

Just out of curiosity, what application are you using a highpass that high and that shallow for?  I'm intrigued  ;D
Logged

slayerizer

  • Recent member
  • *
  • Posts: 6
Re: Bug - high pass filter (v19.0.108)
« Reply #4 on: January 26, 2014, 12:31:05 pm »

   Just out of curiosity, what application are you using a highpass that high and that shallow for?  I'm intrigued  ;D

ahah! I was expecting that one. You often see articles here and there, about 44.1 not being for audiophile and that 96khz or more is the way to go... The Nyquist theorem states that a sampling rate of 44.1khz can reproduce perfectly frequencies up to 22050hz (SR/2). From time to time, when I get my hand on 'high quality' recordings I like to peek here and there to hear what's up there!

My findings, unless proven wrong, is that the recording is always the problem and the medium (CD in that case), is not. I got my hand on digital 192kbps version of vinyl records that sounded way better than CD.

So in general, I use the H-P filter when I'm curious.  ::)  I open the spectrum analyzer and set a range between 21500 and 25k, if I see something, I turn-on the highpass!

I like to peep my frequencies ;D
Logged

slayerizer

  • Recent member
  • *
  • Posts: 6
Re: Bug - high pass filter (v19.0.108)
« Reply #5 on: January 26, 2014, 12:40:00 pm »

Does using a steeper filter slope affect the "failure point"?  I expect it would.

on 44100khz content (without resampling)

a H-P filter of:
 21829hz works for all slopes (6-48db)
 21830hz fails   for all slopes (6-48db)


Logged

dean70

  • Citizen of the Universe
  • *****
  • Posts: 508
Re: Bug - high pass filter (v19.0.108)
« Reply #6 on: January 26, 2014, 03:14:24 pm »

Is it clip protection that is being activated (muting the output)? There was an issue I reported some time ago, where a filter above Nyquist freq would cause clip protection to activate (I have 30k low pass to remove hf noise from some hires & sacd material) - that was fixed, but I wonder if this issue is related.
Logged

Hendrik

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 10935
Re: Bug - high pass filter (v19.0.108)
« Reply #7 on: January 26, 2014, 03:34:50 pm »

Its rather suspicious that it happens at exactly 1% - 220Hz - from the maximum, or 440Hz when doing the same at 88200Hz.
There may be a bit of headroom in the algorithm here for something, maybe I can find out what for.
Logged
~ nevcairiel
~ Author of LAV Filters

mwillems

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 5234
  • "Linux Merit Badge" Recipient
Re: Bug - high pass filter (v19.0.108)
« Reply #8 on: January 26, 2014, 06:22:39 pm »

ahah! I was expecting that one. You often see articles here and there, about 44.1 not being for audiophile and that 96khz or more is the way to go... The Nyquist theorem states that a sampling rate of 44.1khz can reproduce perfectly frequencies up to 22050hz (SR/2). From time to time, when I get my hand on 'high quality' recordings I like to peek here and there to hear what's up there!

My findings, unless proven wrong, is that the recording is always the problem and the medium (CD in that case), is not. I got my hand on digital 192kbps version of vinyl records that sounded way better than CD.

So in general, I use the H-P filter when I'm curious.  ::)  I open the spectrum analyzer and set a range between 21500 and 25k, if I see something, I turn-on the highpass!

I like to peep my frequencies ;D

I get it, you're trying to isolate the high frequency content to see what it sounds like. You might want to try a steeper filter than a 6dB slope.  A 6dB highpass at 20KHz is only 6dB down at 10KHz and only 12 dB down at 5KHz, etc.  So you'll still hear lots of "regular" audio being played with a 6dB high-pass.  When I try a 6dB high pass at 20KHz, I still hear quite a lot of lower frequency content being played back.

By contrast a 48dB filter will be around 24 dB down by 15KHz, which will really isolate the high frequency content.  In fact, if I put a 48dB highpass at 20KHz with most music, I don't hear anything at all, which is not surprising given that I can't really hear much above 16KHz  ;D
Logged
Pages: [1]   Go Up