Hi. After having worked in the field of musicdsp quite some years, there
is still one mysterium left: phase shift. The actual task at hand is to
estimate the frequency-dependent latency of an LTI filter. For a bunch
of filters (i.e. butterworth LP) it works rather well by just calling
Matlab's
As far as causality is concerned it's the *group* delay that should be
non-negative. Phase delay assumes sinusoidal inputs over (-inf, inf) for
which whether a system is causal makes no difference.
-Original Message-
From: music-dsp-boun...@music.columbia.edu
On Mar 17, 2011, at 9:21 AM, Wen X wrote:
As far as causality is concerned it's the *group* delay that should be
non-negative.
well, even group delay is negative with the peaking filters, for
*some* frequencies.
with group delay, there is no issue of phase unwrapping since the
phase
From: music-dsp-boun...@music.columbia.edu
[mailto:music-dsp-boun...@music.columbia.edu] On Behalf Of robert
bristow-johnson
well, even group delay is negative with the peaking filters, for
*some* frequencies.
Yes, but only if the filter has high (negative?) dispersion at that
On Mar 17, 2011, at 12:00 PM, Wen X wrote:
From: music-dsp-boun...@music.columbia.edu
[mailto:music-dsp-boun...@music.columbia.edu] On Behalf Of robert
bristow-johnson
well, even group delay is negative with the peaking filters, for
*some* frequencies.
Yes, but only if the filter
Yes, but only if the filter has high (negative?) dispersion at that
frequency.
i'm not sure what that means. my understanding of dispersion would be
a rapid change of phase or delay vs. frequency.
my understanding is if different frequenices are delayed by different *time*
then you
On Mar 17, 2011, at 2:27 PM, xue wen wrote:
Yes, but only if the filter has high (negative?) dispersion
at that
frequency.
i'm not sure what that means. my understanding of dispersion would
be
a rapid change of phase or delay vs. frequency.
my understanding is if different
my understanding is if different frequenices are delayed by different
*time* then you have dispersion. in this sense the group delay as a
function of frequency can be written as an average group delay
(non-negative if causal) plus a zero-mean item characterizing
dispersion, so that if this item