Question about Brickwall Filter
Hey I just red the following:
"We can achieve a desired slope by combining two filters. For example, we can achieve a 12 dB/oct response by combining two 6 dB/oct filters with the same cut-off frequency. As the signal travels through the first filter, the first octave experiences maximum attenuation of 6dB; as it travels through the second filter, the same octave experiences additional 6 dB of attenuation, resulting in summed response of 12 dB/oct. In all cases, the cut-off frequencies of the two filters should be identical…"
I know that a perfectly vertical brickwall filter is not possible to create, but using this technique I could simply create very steep filter curves just by layering multiple filter object in MSP in a row?
Yes, that is exactly what happens in 2nd-order, 3rd-order, etc. filters. By stacking 6dB/octave filters–all with the same cutoff frequency–you get a steeper and steeper slope by an additional 6dB/octave for each added filter.
Thanks for the reply!
I heard that the number of artifacts in the audio increases with the filter slope. So a higher slope = more artifacts in the audio.
My problem is I don’t understand the point where the artifacts are introduced if every filter attenuates the signal by 6db in a row.
Could you then say that the artifacts in higher filter slopes are the sum of the artifacts that are produced in the single 6db slopes?
Do you have some recommended articles to read on this? I’m just getting some basic knowledge in Audio Technology and would love to dig a little bit deeper into this topic.
"I heard that the number of artifacts in the audio increases with the filter slope"
the phase errors will increase, but the question is if that really matters.
and a 48 dB/A biquad cascade definetly sounds as if it would be a only theoretically possible brickwall filter.
go to filtergraph~ helpfile, look for cascade commands in display mode. very helpful to understand and
have control over what happens when putting filters in series.
in the case of using a Q value > 0., basic rule: the max value of each Q should be limited to 1/number
of filters used in the cascade. (for example Q value 0.5 is "100% resonance" when using 2 6 decibel lowpasses)
Thanks for the help guys!
Just started looking into the helpfile, it’s really good.