1
$\begingroup$

Adobe Audition has a very interesting notch filter that have been very useful to me. It has the following parameters Adobe Audition Notch Filter

According to their documentation, the chosen frequency is the center frequency and the width is either a second, fourth or sixth order, depending on the "Notch Width" option selected.

I want to recreate this same type of notch filter using another tool, e.g. SoX, but I can't understand how it works. I've tried to recreate it using their own "scientific equalizer", using a Butterworth filter with various cutoffs and orders, but could not match the notch filter result.

How does this "Gain" parameter work? How can I reproduce this sort of filter with a traditional bandreject filter, like the one available in SoX?

Improve this question
$\endgroup$
5
  • $\begingroup$ From "center frequency and the width is either a second, fourth or sixth order, depending on the "Notch Width" option selected" it seems to me that the model for this is a simple low-pass filter {LPF} of order 1, 2, or 3 mapped to a band-reject filter (BRF) with a standard mapping in the s-domain and then likely the bilinear transform is used to make it a digital filter. Even assuming all that, there are still unanswered questions that will make your result different from Audition. The Notch Width parameter would be the cutoff of the LPF before mapping to a BRF. It will affect Q. $\endgroup$
    – robert bristow-johnson
    Commented Nov 3, 2023 at 0:23
  • $\begingroup$ The doc can be found here helpx.adobe.com/audition/using/notch-filter-effect.html. Unfortunately it's a weird mix of technical info and marketing babble. It doesn't appear to be a standard notch implementation so without doing some actual measurements, I doubt you can reproduce it . Normally notch filters do NOT have a gain parameter. You can adjust width but the gain at the center frequency is 0 $\endgroup$
    – Hilmar
    Commented Nov 3, 2023 at 1:29
  • $\begingroup$ And what @Hilmar means by "$0$" is $-\infty$ dB. But there is a gain parameter, so this almost sounds like a parametric EQ doing a cut of some specified dB. I am not sure exactly, unless they want a sorta flattened stop band, why they need 4th order or 6th order. Even with 2nd order, you can crank up the Q and have neighboring frequencies be relatively unattenuated. $\endgroup$
    – robert bristow-johnson
    Commented Nov 3, 2023 at 3:07
  • $\begingroup$ @Hilmar Yes, I've noticed that peculiarity regarding the gain parameter on their so called notch filter. Would you please tell me what measurements can I do? $\endgroup$
    – TheGMX
    Commented Nov 3, 2023 at 4:28
  • $\begingroup$ @robertbristow-johnson Yes, one of the experiments I've made was using their own parametric EQ, and although I could get really close using the same gain value and a high Q factor, like 800, for the "Narrow" width option, with their "Very Narrow" setting I could find no correlation between gain and Q factor. At some point I was using much more gain reduction and a Q factor of more than 10k and still could not match their results, i.e., the parametric EQ still affected a larger frequency window. $\endgroup$
    – TheGMX
    Commented Nov 3, 2023 at 4:32

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.