2
$\begingroup$

Consider a FIR audio filter with symmetric impulse response.

In my case, the filter kernel width is 15 frames on each side (so including the center frame, 31 frames total). When examining the audio impulse response of this filter, it includes (A) the initial interval where it approaches the center, then (B) the centered impulse output frame itself, then (C) the subsequent interval following the center.

I have seen numerous terms referring to intervals (A) and (C), including

  • (A) lookahead, ramp-in, processing delay, pre-echo, positive filter width, ring-in, pre-response;
  • (C) decay time, ramp-out, post-delay, negative filter width, ring-out, causal response.

... and many others.

In the audio world, is there a term that is considered most accurate and/or most clearly descriptive for these two intervals?

Improve this question
$\endgroup$

1 Answer 1

Reset to default
2
$\begingroup$

That's the causal (comes after the impulse) and anticausal part.

Improve this answer
$\endgroup$
4
  • 1
    $\begingroup$ Those seem accurate of course. In referring to the intervals themselves (perhaps comparing their lengths to those of another filter), you'd talk about the "anticausal response length" and the "causal response length"? I guess I was hoping for something crisp like ... group delay? $\endgroup$
    – MGP
    Commented Nov 10, 2020 at 19:39
  • 1
    $\begingroup$ "Group delay" is a property of the overall filter however (in case of your symmetric filter it's half the filter order, by the way, always). So, yeah, if you just want to describe the length of these intervals instead of the intervals themselves, then "group delay" would be appropriate (but only for linear phase filters like your symmetric ones!). $\endgroup$
    – Marcus Müller
    Commented Nov 10, 2020 at 19:44
  • $\begingroup$ I see. So for a filter that is not symmetric, is it accurate to refer to the length of its anticausal response as group delay? $\endgroup$
    – MGP
    Commented Nov 10, 2020 at 19:49
  • 1
    $\begingroup$ @MGP no. Group delay is then a frequency-dependent property of the overall filter, and the length of the anticausal or causal parts don't tell us anything about it. $\endgroup$
    – Marcus Müller
    Commented Nov 10, 2020 at 19:54

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.