So I understand it is possible to display the frequency response of normal filter through ploting the magnitude with its biquad coefficients…
However, how in the heck do you plot the frequency response to a comb-filter since it is created using one sample of delay to my understanding?
Maybe a dummy buffer with only 1’s or something? Then take the FFT of the dummy buffer? Or is it simply just using the two coefficients and utilizing the visual path’s delay?
If anyone has direction I will go to it!
Live Hack of the day: play an impulse plus a bunch of zeros through a copy of the filter and use an FFT to get the magnitude plot. You can run the whole operation at a lower “sample rate” for better performance but the further you reduce, the more low end resolution you will loose.
This - of course - works on any filter and is especially useful if calculating the frequency response of your filter is challenging, for example if it is some esoteric analog model.
2 Likes
The method with the impulse is great because it works with any filter / system, but a comb filter also has a transfer function, just like a biquad (but it has a different form). So you can also evaluate this transfer function to get the frequency response. Or even do both and compare them.
2 Likes
Exponential functions f(t) = c**t are the eigen-functions for time-invariant linear operators, aka filters. The respective eigen-value for the eigen-function with c = e**(i*r) would represent the response for the frequency r in radians.
1 Like
