[SoX-users] find clicks

[Erich Eckner]

Dear list,

probably this is beyond sox' scope, but maybe one of you can help anyway:

I have some recordings (flac-files) which have emp-interference (audible
"clicks" of length ~ 1/2000s) in them. Do you know any scriptable tool
(or a plugin to sox) to extract the positions (time or sample) of
potential clicks?

I know, I could track them down manually, but I would really prefer to
listen only to a few minutes of potentially emp'ed audio rather than to
many days of audio (I usually do multi channel recordings and the emp is
not on all channels and also very sparse on the time axis).

Furthermore (this is probably closer to what sox can do), is there a
effect combination in sox (or a ladspa plugin) which can "iron out"
these disturbed samples (and intrapolate their values somehow from
surrounding samples) - given, I know the position of the samples?

regards,
Erich

[Peter P.]

Hi Erich,

Post by Erich Eckner
Dear list,
I have some recordings (flac-files) which have emp-interference (audible
"clicks" of length ~ 1/2000s) in them. Do you know any scriptable tool
(or a plugin to sox) to extract the positions (time or sample) of
potential clicks?
I know, I could track them down manually, but I would really prefer to
listen only to a few minutes of potentially emp'ed audio rather than to
many days of audio (I usually do multi channel recordings and the emp is
not on all channels and also very sparse on the time axis).
Furthermore (this is probably closer to what sox can do), is there a
effect combination in sox (or a ladspa plugin) which can "iron out"
these disturbed samples (and intrapolate their values somehow from
surrounding samples) - given, I know the position of the samples?

A quick web research on "ladspa declick" gave the following results:

Not SoX but Audacity has Click Removal and Repair effects.

Then this link
http://home.snafu.de/wahlm/dl8hbs/declick.html

and not sure if gnoise
https://sourceforge.net/projects/gnoise/
does something.

tell us how things go
P

[Erich Eckner]

Hi P,

thanks for your reply.

*ouch* I totally forgot about audacity - I regularly use this to "look
at" the audio (while I process it with sox) ...
A first trial of audacities "Click removal" removes most of the clicks,
however, it also alters all audio where no clicks have been - this seems
to be intrinsic to an analogue-click removal tool.

declick only seems to work with "digital" clicks, e.g. single samples
being wrong (according to its website) - wheras I have ~50 samples
forming a "nice" dip.

and gnoise unfortunately doesn't compile on my machine, because it can't
find gtk - I have gtk2 and gtk3 installed (and gtk is not available on
arch linux), so I think it's just outdated :-/

I think, I'll experiment some further with audacity and probably write
my own tool at some point which conforms to my crazy opinions standard,
e.g: no change of audio, where no click is suspected, and removal of all
suspected clicks.

cheers,
Erich

Post by Peter P.
Hi Erich,

Post by Erich Eckner
Dear list,
I have some recordings (flac-files) which have emp-interference (audible
"clicks" of length ~ 1/2000s) in them. Do you know any scriptable tool
(or a plugin to sox) to extract the positions (time or sample) of
potential clicks?
I know, I could track them down manually, but I would really prefer to
listen only to a few minutes of potentially emp'ed audio rather than to
many days of audio (I usually do multi channel recordings and the emp is
not on all channels and also very sparse on the time axis).
Furthermore (this is probably closer to what sox can do), is there a
effect combination in sox (or a ladspa plugin) which can "iron out"
these disturbed samples (and intrapolate their values somehow from
surrounding samples) - given, I know the position of the samples?

Not SoX but Audacity has Click Removal and Repair effects.
Then this link
http://home.snafu.de/wahlm/dl8hbs/declick.html
and not sure if gnoise
https://sourceforge.net/projects/gnoise/
does something.
tell us how things go
P
------------------------------------------------------------------------------
Developer Access Program for Intel Xeon Phi Processors
Access to Intel Xeon Phi processor-based developer platforms.
With one year of Intel Parallel Studio XE.
Training and support from Colfax.
Order your platform today.http://sdm.link/xeonphi
_______________________________________________
Sox-users mailing list
https://lists.sourceforge.net/lists/listinfo/sox-users

[fmiser]

Post by Erich Eckner
probably this is beyond sox' scope, but maybe one of you can
I have some recordings (flac-files) which have emp-interference
(audible "clicks" of length ~ 1/2000s) in them. Do you know any
scriptable tool (or a plugin to sox) to extract the positions
(time or sample) of potential clicks?
Furthermore (this is probably closer to what sox can do), is there a
effect combination in sox (or a ladspa plugin) which can "iron out"
these disturbed samples (and intrapolate their values somehow from
surrounding samples) - given, I know the position of the samples?

Gramophile? I use it for de-clicking vinyl LP recordings.

[Erich Eckner]

Hi fmiser,

thanks, this has exactly the effects, I was looking for.
Just for the archive: It's 'gramofile'.
It won't process my 24bit-wavs, but I'll get around that.

regards,
Erich

Post by fmiser

Post by Erich Eckner
probably this is beyond sox' scope, but maybe one of you can
I have some recordings (flac-files) which have emp-interference
(audible "clicks" of length ~ 1/2000s) in them. Do you know any
scriptable tool (or a plugin to sox) to extract the positions
(time or sample) of potential clicks?
Furthermore (this is probably closer to what sox can do), is there a
effect combination in sox (or a ladspa plugin) which can "iron out"
these disturbed samples (and intrapolate their values somehow from
surrounding samples) - given, I know the position of the samples?

Gramophile? I use it for de-clicking vinyl LP recordings.
------------------------------------------------------------------------------
Developer Access Program for Intel Xeon Phi Processors
Access to Intel Xeon Phi processor-based developer platforms.
With one year of Intel Parallel Studio XE.
Training and support from Colfax.
Order your platform today.http://sdm.link/xeonphi
_______________________________________________
Sox-users mailing list
https://lists.sourceforge.net/lists/listinfo/sox-users

[fmiser]

Post by Erich Eckner

Post by fmiser

Post by Erich Eckner
I have some recordings (flac-files) which have
emp-interference (audible "clicks" of length ~ 1/2000s) in
them. Do you know any scriptable tool (or a plugin to sox) to
extract the positions (time or sample) of potential clicks?

Gramophile? I use it for de-clicking vinyl LP recordings.

thanks, this has exactly the effects, I was looking for.
Just for the archive: It's 'gramofile'.
It won't process my 24bit-wavs, but I'll get around that.

Your welcome. As to the name, I guess this is a downside to
tab-completion and a weak memory. I type "gramo"-TAB. *grin*
Thanks for the correction.

Then 'cause I'm curious - how do you plan to get around the 24-bit
problem?

[Erich Eckner]

Post by fmiser

Post by Erich Eckner

Post by fmiser

Post by Erich Eckner
I have some recordings (flac-files) which have
emp-interference (audible "clicks" of length ~ 1/2000s) in
them. Do you know any scriptable tool (or a plugin to sox) to
extract the positions (time or sample) of potential clicks?

Gramophile? I use it for de-clicking vinyl LP recordings.

thanks, this has exactly the effects, I was looking for.
Just for the archive: It's 'gramofile'.
It won't process my 24bit-wavs, but I'll get around that.

Your welcome. As to the name, I guess this is a downside to
tab-completion and a weak memory. I type "gramo"-TAB. *grin*
Thanks for the correction.
Then 'cause I'm curious - how do you plan to get around the 24-bit
problem?

I'll try two things:
1. fast and ungly: try to fiddle larger/unecpected samples into the
chain (disable whatever check causes gramofile to exit, pad by zero
bits, replace ints by long ints, etc.)
2. slow and nice: try to copy the actual algorithm into a standalone program

cheers,
Erich