has anyone made a click remover in pd?
care to share the patch?
mine is not working for some reason.
yes, i could also just disconnect the dac~ so that i don't hear any clicks.
but actually what i want is a click remover. - something that will leave a complex audio signal fairly intact but remove the clicks caused by discontinuities.
Hey, Derek (right?)
I made a weighted median filter abstraction (not good) and an external (not finished), a while back. I'll finish the external up, quickly, test it and send you the code.
As an example of what it does, here's the abstraction so you can see it. The abstraction is just too cpu-hungry to be practical. Array1 has a signal, corrupted by brief (1 sample) clicks. Array2 is the filtered signal.
Alright... more later... Chuck
On Wed, Jan 7, 2009 at 3:00 PM, hard off hard.off@gmail.com wrote:
yes, i could also just disconnect the dac~ so that i don't hear any clicks.
but actually what i want is a click remover. - something that will leave a complex audio signal fairly intact but remove the clicks caused by discontinuities.
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Wait... you said discontinuities. I've got something that takes out impulses...
A low-pass filter should do a reasonable smoothing, but it would take a lot away from the signal, in the other spots. What if... you could detect the discontinuities using Hilbert~. I just started playing around with that. I could see the discontinuities int a Hilbert~/atan2~ pair in a test. If you use a threshold on that, you could ramp in the low-pass filter in the places you want...
Chuck
On Wed, Jan 7, 2009 at 3:00 PM, hard off hard.off@gmail.com wrote:
yes, i could also just disconnect the dac~ so that i don't hear any clicks.
but actually what i want is a click remover. - something that will leave a complex audio signal fairly intact but remove the clicks caused by discontinuities.
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
On Wed, 7 Jan 2009, Charles Henry wrote:
Wait... you said discontinuities. I've got something that takes out impulses... A low-pass filter should do a reasonable smoothing, but it would take a lot away from the signal, in the other spots.
In the discrete domain, a loud very-high-frequency tone is indistinguishable from a sequence of discontinuities.
So, what is it that people are calling a "discontinuity" in this thread?
If discontinuities are so hard to remove, it's because they don't correspond to anything in objective reality. It's all in the mind.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
click remover in the "practical sense" of the meaning work quite
well. as practical meaning I refer to what is written here
http://audacityteam.org/wiki/index.php?title=Click_Removal
10 years ago we spent fortunes for cedar hardware processors (which
did and do an incredible good job). what is difficult is the
difference between "wanted" transients" like very short sounding
percussion and "discontinuities" like unexpected jumps in level,
waveform etc. In order to make the difference it takes complex
analysis processes that determine tempo so that "musically logical
discontinuities" are untouched - even if they're not metronomically
precise.
Low-pass cannot do the job because the "click" is typically broadband
- but you're right, we may perceive a click as high frequency. In
this regard, the process would identify a 20 kHz saw tooth as
musically correct because it is periodic. clicks are not.
Best Jurgen
On Jan 8, 2009, at 10:31 AM, Mathieu Bouchard wrote:
On Wed, 7 Jan 2009, Charles Henry wrote:
Wait... you said discontinuities. I've got something that takes
out impulses... A low-pass filter should do a reasonable smoothing, but it would take a lot away from the signal, in the other spots.In the discrete domain, a loud very-high-frequency tone is
indistinguishable from a sequence of discontinuities.So, what is it that people are calling a "discontinuity" in this
thread?If discontinuities are so hard to remove, it's because they don't
correspond to anything in objective reality. It's all in the mind._ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal,
Québec_______________________________________________ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
-
Charles Henry -
hard off -
Lao Yu -
Mathieu Bouchard -
mescalinum＠gmail.com