Groove-y: Playing Vinyl Records Optically

Fellow eFlea (electronic Flea Market) aficionado and vintage audio nut Steve Williams sent along some interesting links to various sites that describe playing old vinyl records with optical techniques. Steve writes:

[Here is a] European group trying to "go commercial"
http://www.jts2004.org/english/proceedings/Stotzer.htm
http://www.eif.ch/visualaudio/

I don't think these guys are related to the people in Berkeley, Dr. Haber etc. The euro. group above sounds like it is trying to take the process commercial. They are also using another antique process - photographic film -rather than microscopy techniques such as the UC Berkeley boys.
http://sciencematters.berkeley.edu/archives/volume4/issue30/story1.php

The first attempts I ever heard of were in around 2000 when people proposed using "high res." scanner technology to extract the image like this example:
http://www.cs.huji.ac.il/~springer/
or
http://saturn.vfx.com/mccann/visionfinal/doc/

So, not counting the variation of the laser turntable, which has been developed for a long time, there are now a fair number of people working on this.

I can't help but notice however that in all of them (except the laser player) there is always a resolution issue leading to the inability - so far at least - to resolve the HF info of even a mono wide groove 78 disc. This, if it is referred to at all, is euphemistically and optimistically called a "low pass filter" to "eliminate noise". [Update-- Steve later noted: Actually I think STM was what I was thinking of. And I take back a bit of what I said. The folks at Berkeley ARE using confocal - now. Their latest efforts with their dedicated gen-2 machine do have acceptable HF response.]

The sound samples from the euro group, to my ear have nothing above about 5-6KHz.
If you are archiving even spoken word from 19th century source material it would be nice to have some consonants rather than all vowels so that one might understand what they were saying. What is needed is a methodology that will resolve every bit of HF that is present and better techniques in the determination and differentiation between what little HF information there is lying down on the noise floor and the noise itself. The present state of the art in optical extraction of sound waves still seems to amount to little more advantage than putting a giant capacitor across the audio, other than the obvious fact of being able to play things that are broken or considered too valuable to play by means of stylus contact.

So, Dave [Ruigh], what is the methodology you keep talking about for accurate surface detection? Does it involve physical contact? Is it able to resolve something as small as you have stated the smallest features of a stereo microgroove record contain? Would it have to "play" the disc rotationally as in conventional stylus (or laser) play or could it be used to "scan" obtaining the advantage of these other methodologies in reconstructing damaged surfaces?

I want to see resolution down to, say, recovery of the front minus rear modulations of the CD-4 carrier!

Dave Ruigh responded with these thoughts:

I think the tech Steve is alluding to is an optical machine known as a confocal microscope. I believe this is what the Berkeley folks are using. There is also another unrelated technology known as AFM (I think) Atomic Force Microscopy, and it's variants, which include the STM (scanning, tunneling), and one or two others I've forgotten about. No AFM could work real-time, and I think the CFM scans records at something like 0.01X.

Computers, scanners and software do some things really well; playing old records isn't one of them.

The STM is an older version of this class of surface scanning instruments; AFM is actually a newer invention (according to Wikipedia). AFMs are capable of extracting all the high frequency information you could ever want, with resolution below a nanometer. You could expect a scan rate of about 1 record a decade.

Into this mix I will toss my idea of getting better signal to noise from old scratchy records. It is well known that you can parallel amplifiers and reduce noise, since the signal contribution of the amps is arithmetic but the noise adds RMS. Similarly if you had ten records all pressed from the same metal master you could phase align the audio and add all ten together, and the pops clicks and other noise would also add RMS. You may need an encoder on the turntable to precisely record phase, but those records are pressed from the same master and they will stay in good phase relations, perhaps even up to high audio frequencies.[Update:] Former EDN editor-in-chief Steve Leibson supplied this link to a company that used to rep a Japanese laser turntable.


Source [EDN]