I have a pre-designed in-series 14th-order IIR filter in second-order sections. Problem is, when I simulate and measure the behavior in my codec, there is a frequency shift downward (as compared to the MATLAB prediction). Is there a simple way to alter the coefficients in such a way to shift the center frequencies of each band? Thank you.
$\begingroup$
$\endgroup$
-
$\begingroup$ Do you need to shift the frequencies or scale them by a factor? $\endgroup$ – Olli Niemitalo Sep 14 '15 at 15:24
-
$\begingroup$ Is this the same problem as in this question? dsp.stackexchange.com/questions/19225/… $\endgroup$ – JRE Sep 14 '15 at 15:27
-
$\begingroup$ JRE, I don't think so. The 14th-order SOS filter I've been provided with does not use RBJ. @Olli, I don't know, I can't tell if the shift is necessarily uniform from 0 to Fs/2. I'm trying it by hand first. $\endgroup$ – panthyon Sep 14 '15 at 15:36
-
$\begingroup$ OK, but could it still be subject to the kind of warping that was mentioned, with maybe a similar solution? $\endgroup$ – JRE Sep 14 '15 at 15:46
$\begingroup$
$\endgroup$
The problem was an incorrect sampling rate issue, as a lower sampling rate than the designed rate caused the center frequency shift downwards. Today I learned something about the relationship between sampling rate and IIR coefficients (there is a proportional relationship between sampling rate and center frequency).
