Building on Tom's idea of optical scanning, couldn't you sample the
groove wall at three or four different heights and use a computer to
compare them to take out areas interpreted as noise and combine areas
that contain a music signal? Except for pits, digs, or deep
scratches, the noise would be random up and down the side of the
On Sat, Nov 24, 2012 at 10:09 AM, Tom Fine <[log in to unmask]> wrote:
> Hi Doug:
> I was thinking about this concept. Do you think you'd get a well-defined
> enough negative of the groove to play it back? It seems to me that you're so
> many generations removed from the original groove that what you end up with
> will be too physically different and distorted to sound any good. I thought
> one of the problems with shellac is that it ends up with a non "sharp"
> reproduction of the groove. Plus, wouldn't you be "baking in" surface noise
> from the non-smooth shellac surface? I was thinking using optical scanning
> and "photoshopping" the resulting image, you could sharpen up the groove's
> pitch and path and also "erase" the roughness of the walls and bottom.
> -- Tom Fine
> ----- Original Message ----- From: "Doug Pomeroy" <[log in to unmask]>
> To: <[log in to unmask]>
> Sent: Friday, November 23, 2012 10:53 AM
> Subject: Re: [ARSCLIST] Stuff of which dreams are made
> Indeed, a machine cannot create out of thin air something it "thinks" is
> "missing", but
> the technique of interpolation (now more than twenty years old) can often
> provide a
> convincing fill-in. I believe Sonic Solutions' NoNoise was the first,
> followed by processors
> from CEDAR and many others using versions of the basic idea: analysis of the
> before and after an audio disruption, be it a "click" or scrape or whatever,
> can give the
> computer enough information to synthesize something like the missing music.
> doesn't always work, of course, especially if the texture of the music is
> rapidly changing,
> or if the duration of the disruption is very long.
> The device (IRENE) which Carl Haber developed is a unique tool, originally
> for playing vertical recordings, but the audio quality has, so far, been
> less than ideal.
> One idea which might prove fruitful, would be to make a new negative
> impression from
> a worn 78, and plate it and play the plated metal part with the bi-point
> styli designed for
> playing negatives. The idea here is that the ridge on the negative
> represents the very
> bottom of the original groove, and since most groove wear is on the sides
> and not at
> the very bottom, this might allow a cleaner transfer of the music.
>> Date: Thu, 22 Nov 2012 11:53:22 -0500
>> From: Steve Smolian <[log in to unmask]>
>> Subject: Re: Stuff of which dreams are made
>> Seems to me this is a two-parter.
>> The first is to rebuild one wall from the attributes of the other.
>> The deeper problems occur when both walls are so damaged that the
>> issue enters. In many cases, looking for an identical iteration of a note
>> or phrase from an un- or lesser damaged segment elsewhere, particularly in
>> the "A" of songs in "AABA" form, would be the next step. After that,
>> synthesis has to take over. Of course, an second copy, no matter how
>> damaged, could enter the picture. If done with the electron microscope,
>> needn't even be trackable.
>> As the late Mr. Saul said, "The Future Lies Ahead."
>> Steve Smolian
>> -----Original Message----- From: Tom Fine
>> Sent: Thursday, November 22, 2012 10:43 AM
>> To: [log in to unmask]
>> Subject: Re: [ARSCLIST] Stuff of which dreams are made
>> The only comment I have on this line of thinking is, it's very hard for a
>> machine to create out of
>> thin air something it "thinks" is "missing." When you get into something
>> involving human senses and
>> brain together -- music, art, photography, moving pictures -- it's not
>> an "artificial
>> intelligence" thing but also an "artificial aesthetic" thing.
> Doug Pomeroy
> Audio Restoration & Mastering Services
> Transfers of metal parts, lacquers,
> shellac and vinyl discs & tapes.
> 193 Baltic St
> Brooklyn, NY 11201-6173
> (718) 855-2650
> [log in to unmask]