De-Wowing an entire LP...

Has anyone tried this before?

We've been working on an LP of Quintetto Chigiano's Franck Piano Quintet, one of the first LPs UK Decca ever released, and due for reissue on 1st March.

In my opinion it is a beautiful recording, yet unlike I think all the other Quintetto Chigiano recordings, was never reissued by Decca following its 1950 release. When we came to start work on it we realised immediately why not - the recording suffers from a continual pitch 'wow' every 0.9 seconds. Closer inspection revealed this to be a product of the master recording (you can spot edits in FSE where the pitch suddenly jumps rather than smoothly varies). The result is a piano with nasty vibrato and particularly wayward strings - something to induce seasickness after a few bars...

Well, thanks to the tenacity and doggedness of my colleague Peter Harrison, this wonderful recording is to see the light of day again, sounding like it never has before. He's gone into it and, wow by wow, every 9/10ths of a second for over 36 minutes, smoothed out every single one using digital pitch-bending in AA1.5. This is very, very tricky to get right and it's very, very easy to make it worse. 2,400 wows later we have a beautiful rendition of the Franck Piano Quintet and I'm personally in awe of what Peter's achieved.

We're not sure about this, but this may just be the first example of an entire LP being digitally de-wowed. Anybody know of any others?

What is curious in this case is that the problem was constant but at double the rotational speed of the LP - e.g. 66.6 times per minute. We've yet to figure out exactly how this came to pass, but are sure it was a fault on the original master - any ideas as to how it might have got there?

Wouldn't it be great if you had a program where you just type
adjust every 0.9 second by x amount. Or even just say it in a mic.

Wouldn't it be great if you had a program where you just type
adjust every 0.9 second by x amount. Or even just say it in a mic.

We investigated this - but of course the 0.9s itself varies slightly, coming from an analogue source. Because of this and of the edits which interrupt the wow and reset its position, meant that at best Peter could only do a few at a time. The hardest points to get right were the in and out points, apparently, so a 9s de-wow master preset turned out to work better than ten 0.9s presets applies one after the other...

Wouldn't it be great if you had a program where you just type
adjust every 0.9 second by x amount. Or even just say it in a mic.

This shows you have very little understanding of the problems involved in an exercise of this nature.  Automatic de-wowing is not something I expect to see in my lifetime .  Actually, having said that, I do know one our forum members was working on something like this - but I've heard nothing about it for some considerable time now, so I imagine he has abandoned it.

To answer the original question, I'm not aware of a complete LP having been treated in this fashion, although I do know of a couple of 45 singles done by someone on the USA.  Personally, I don't think I would have had the patience (never mind 'tenacity and doggedness') to do something like this - OK, if the cheque was big enough, I suppose I could be persuaded - and I take my hat off to anyone who has.  

In the past, I have sorted out some stuff which was cut on a lathe where the cutting table slowed down as the recording progressed - and that was a PITA of nothing like the same dimension.

What is curious in this case is that the problem was constant but at double the rotational speed of the LP

Going to say something stupid: half-speed mastering? But it seems early for that.

I have terrible espect for someone who can do that. I would have given up .

One wild and whacky thought which occured to us involves - at some point early in the recording process - a master recording made direct to disc at 66.667 rpm - possible an adapted 78rpm cutting lathe slowed down and rejigged for microgroove recordings, but screwed up in the process such that it produced the effect here.

Reasons to think this?

1 - The disc itself is absolutely dead centred

2 - The wow corresponds to an off-centre LP at double-speed (i.e. 66rpm)

3 - It's been edited - therefore transferred onto tape - after the wow was introduced.

Could this be an early, post 78rpm ffrr, Decca 33rpm experiment which made it out?

What is the speed of a capstan in a tape machine?

What is the speed of a capstan in a tape machine?

Nowhere near that slow.  I wondered about drag from a tape supply spool, but that would vary as the tape diameter reduced.

Paul

What is the speed of a capstan in a tape machine?

Like Paul says, faster than that! If you assume that the capstan is 1/2" diameter, then it's got a 1.57" circumference. If the tape was travelling at 15ips, that would give a cyclic frequency of 15/1.57 = 9.55Hz, or just over 19Hz at 30ips.

Wouldn't it be great if you had a program where you just type
adjust every 0.9 second by x amount. Or even just say it in a mic.

This shows you have very little understanding of the problems involved in an exercise of this nature.  Automatic de-wowing is not something I expect to see in my lifetime .  

All I said was wouldn't it be great.........
It's like saying wouldn't it be great if the world was chocalate!  

I've been discussing this elsewhere and received this amongst the replies:

"...there is an engineer in California named Dimitrios Antsos who developed such a program.  He is able to successfully "de-wow" recordings which had been dubbed from off-center sources (such as some AFRS radio transcription discs), and more recently has done some restorations for Music & Arts CDs.

Mark Obert-Thorn"

However, a search on this revealed a less-than-flattering review from 2001 - one can only hope the program has been tweaked since:

"...many of the sides are affected by "wow," a variation in speed that resembles the effects of seasickness. Supposedly this has been corrected with a special computer program created by one Dr. Dimitrios Antsos. If this release is "corrected," the uncorrected version must be truly appalling."

(from: http://www.classical.net/music/recs/reviews/m/m&a01085a.html)

I hope later today to have some before and after audio samples and some FSE screenshots up on our site - I'll post a link here when they're ready, for those who're interested...

Dimitrios Antsos.... He is Greek. God bless him.

I've put example files and FSE screenshots online on a partially constructed page here:

http://www.pristineaudiodirect.com/LargeWorks/Chamber/PACM023.php

Have a look and listen...

Actually, having said that, I do know one our forum members was working on something like this - but I've heard nothing about it for some considerable time now, so I imagine he has abandoned it.

Um, that would be me.

Since he has come out of the closet, I confirm it was to younglove that I was referring .

Obviously, for most of us, the usual problem we have is a single source recording (with the wow) so his solution isn't going to help with that.  That said, it's an interesting idea.  

Are you going to make this application available for us to play around?