Hello,

This may not be the right forum for this, but I wanted to run something by
the list to see if my logic is correct.  Suppose I have been given a 9-point
impulse response for filtering by partitioned convolution, and this impulse
response has been worked out so that its frequency response is optimal for
sr=44100.  If I want to be able to apply the same filter in other sample
rates so that it has the same real frequency response (i.e. "sounds the
same"), will I be able to just resample the original impulse response?  If
I'm using a sinc interpolator to resample, and I get several hundred points
in the new response, will that be a problem?  My reasoning is this: if I
filtered a 44100 source with the 9-point filter and then resampled it,
should it be the same as if I resampled the source and the IR separately and
then convolved after?  If so, shouldn't the resampled IR be correct for
filtering in other sample rates?

On the other hand I feel like if the original impulse response was
calculated to effect a specific pole/zero distribution, should I be able to
just move the zeros around and get a new (presumably shorter) impulse
response with a similar frequency response?  I don't know enough of the math
yet to know how to do this, but just resampling the IR and going with it
seems like voodoo.

Thanks,

Matt