I saw one of the demos that Haber and his associate did in New York.
I think that an image-based approach to noise reduction is something that 
ought to be explored.
From the images I saw at the demonstration, it is clearly obvious to me 
which parts of the image are the walls of the groove, and which parts were 
imperfections in the groove wall.
If the imperfections were "healed" before the images were converted to audio 
signals, that would create a more faithful reproduction of the original 
recording, no?
Quite a labor-intensive task, though. Having to correct each image 
individually. I suppose algorithms would have to be developed..
-Matt Sohn
P.S. Is my reply-to thing coming through correctly? I am composing through 
Windows Live, instead of Yahoo.

-----Original Message----- 
From: Doug Pomeroy
Sent: Saturday, October 26, 2013 12:55 PM
To: [log in to unmask]
Subject: Re: [ARSCLIST] Advice needed on removing / minimizing tape 
bleed-through

Hi Tom,

I wish I could share your optimism about future developments.

Haber's work has been concerned with developing and improving his optical 
mechanism.
As far as I know, he hasn't done any work on audio restoration (the first 
demo which he
released several years ago was quite noisy), as it is not his field of 
expertise.  But the ultra
high resolution of his scans offer much information about the nature of disc 
surface noises,
which may help those working to perfect noise removal processors.

Doug Pomeroy
[log in to unmask]


> Date:    Fri, 25 Oct 2013 08:18:52 -0400
> From:    Tom Fine <[log in to unmask]>
> Subject: Re: Advice needed on removing / minimizing tape bleed-through
>
> Hi Ted:
>
> Much better!
>
> The wonders of modern DSP in careful craftspeople's hands!
>
> The next frontier will be figuring out how to grab the just musical 
> content out of a noisy grooved
> disk and then un-do the problems of groove distortion and disk wear. I 
> hope Carl Haber's work leads
> there -- scan the groove and then come up with some kind of Photoshop-like 
> algorhythm to "heal"
> groove wear and the material on the groove surface that produces playback 
> noise (I'm assuming that
> comes down to rough-surface shellac, which would need to be differentiated 
> from minute lateral
> changes in the groove, ie soft-dynamic music content). I am optimistic 
> that such a system will
> emerge in my lifetime. Imagine few-dozen-dollar software that enables you 
> to scan your 78RPM disks
> on a high-resolution flatbed, then "heals" the ravages of time and the 
> problems with the original
> shellac compound and saves a clean,crisp audio file to your hard drive.
>
> -- Tom Fine