[Table of Contents]


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: arsclist Thoughts on a "dream" manual declicker/deglitcher for 78 restoration (Windows-based software)



James wrote:
>Jon Noring wrote:

>> It is entirely possible that somebody has built the tool described
>> here: a manual declicker using frequency-space for visual location
>> of clicks/glitches, and a robust removal and reconstruction
>> algorithm to rebuild the wave form at the point the glitch was
>> removed.

> I would agree with Graham that the CEDAR tools are not expensive if you
> are restoring audio professionally. Consider how much your time costs
> and then consider how much time is saved by using the right tool.

Obviously if one has a full-time business to do audio restoration one
should seriously consider something like CEDAR or NoNoise. I believe
the NoNoise tools are now being (or will soon be) sold through ProTools
(or whatever) for about $2000. Not bad, actually. I just need to buy
a Mac. <laugh/>


> However, for your purposes I would suggest that you look at Cool Edit
> Pro. This has the spectral view which makes it easy to find clicks and
> then a select, button click to bring up dialogue box and button click to
> remove glitch process for each glitch.

I have tried Cool Edit, and found it very wanting. The three step
process to remove clicks took way too long, plus the algorithm it uses
to "subdue" the click is pretty useless -- in many cases it actually
created artifacts. Why it can't simply cut it out and then reconstruct
it using the information on both sides of the cutout portion is beyond
me -- in most cases it'd probably be a more accurate restoration.

Anyway, with Pristine Sounds 2000, which is conceptually very close to
what I've been talking about, I can manually remove glitches at a rate
of one every 3-5 seconds. Some of the really difficult three minute
recordings may have over 1000 glitches to remove, even after gently
applying automated declicking/decrackling (I've found that if one
stomps on these automated tools, they really distort the sound.)


> 2 or 3 years ago I tried the demo versions of just about all the glitch
> removal software that I could find (except the really expensive
> hardware/software) and I felt, like you, that most of it was cumbersome
> to use. For most basic noise reduction purposes I found that Cool Edit
> Pro worked as well as anything for the occasional transfer that I have
> to do.

If you haven't, I'd suggest you try the Pristine Sounds 2000 demo.
The demo can be downloaded from:

http://www.alienconnections.com/downloads.htm

(Look for Pristine Sounds 2000, build 00.07.26)

Glitch removal with PS 2000 is done by first opening the wave you wish
to edit. Then in the Function menu item select "Frequency Space Edit..."
A gray sonogram window comes up. I found that for click removal I set
the brush tool to "Declick brush", 99% strength and as thin and long
(high) as the settings allow. Then place the cross-hairs cursor on top
of a click (which is a thin vertical line, sometimes which only partly
covers the frequency range -- note that not all thin vertical lines
are noise! -- they can be sibilants of some sort -- a listening test
will verify if the vertical line should be there or not.)

Note that the demo will only save waves which are 30 seconds or shorter.

Now, to comment on your mention of time. Considering that I've found
the number one time sink for restoring old recordings is manual glitch
removal, to clean up what the automatic declickers and decracklers do
not remove (and you don't want to stomp down hard on these automated
tools to do 100% removal), the tool I advocate would significantly
reduce the time to do this, even for toolsets such as CEDAR and
NoNoise which I understand do not use a one-step sonogram approach as
PS 2000 does. It should also be more thorough and catch clicks that
otherwise will escape. I've ran through PS 2000 the work of a couple
professional restorers I know and immediately caught a few clicks they
missed, and which when you listen to a couple times can hear the
clicks, albeit very subtle. By visualizing using a sonogram, you can
see everything clearly -- here the eye will see and locate clicks much
better than the ear can (not to say the ear is not important!)

Jon Noring

-
For subscription instructions, see the ARSC home page
http://www.arsc-audio.org/arsclist.html
Copyright of individual posting is owned by the author of the posting and
permission to re-transmit or publish a post must be secured
from the author of the post.


[Subject index] [Index for current month] [Table of Contents]