Let's leave Hank...in the shadows?

[jay]

I was catching up on my reading of old posts and listening to the sound files created by Charlie and in particular to the 362436 track which demonstrates the H&C echo unit prototype.He mentioned the following:-

"The backing track is derived from the Shadows original stereo recording. I processed it to remove Hank as much as possible at all times by reversing the phase of one channel and summing the two tracks to mono. I then added a simulated stereo effect. There is no stereo spread of instruments as such but there is sufficient clearing of the centre part of the image for my new lead guitar part to fit into. Some parts of Hank remained in the stops, so I reduced the level as much as I possibly could, without removing drum and other sounds that had to remain. The only part that remains and cannot be removed is the reverb that was increased by the studio at the ending fade out. Other than that reverb at the ending fade, any reverb you think you are hearing isn't reverb at all, it is from the echo unit that I used."

I am interested in picking out some of those early Shadows tracks which are susceptible to this treatment although as yet I don't have the latest software to do the job. I understand that not all of the tracks yield good results....

I'm sure somebody has had success with this other than Charlie and maybe they can give us a few pointers as well.

What software programmes are being used....are some better than others? Is it only centre vocal cancellation which is on offer...which seems to also adversely affect the other instruments. I've heard one or two examples of this and the quality leaves a lot to be desired. But I've a YT sample that someone did (not of Hank but of Mark Knopfler from his DS days) which is remarkable as they took Money For Nothing and totally isolated his guitar and his second guitar overdubs as well and there is no trace of any vocal or even of any other instrument. Wonder how they did that? Could they have gained access to master tapes somehow.....

Curiously there is one Shadows track which doesn't need any software to remove Hank and I've been playing along to this for years...but I've not found any others. Has anyone?

jay

[Charlie Hall]

Hi Jay,
As far as I remember with Shindig it was simply a case of removing the right side track (which consists of Hank alone!) and replacing it with my take. Is this the tune you are talking about? Others are not so easy and I think impossible in some cases without losing too much other content. With almost all of them extra tricks are needed to eg. to balance out the bass etc. All I basically do is put a track in Sonar and process it from there with phase cancelling and summing to mono.
Paul Rossiter seems to have had success with removing Hank more effectively than I was able to, perhaps he could give some clues.
Regards,
Charlie

[jay]

Hi Charlie........yes Shindig is another one but the one I was thinking about was The Breeze and I. One day I just turned the stereo balance control all the way to the left and Hank stopped playing...magic. All right he's still there but so far away he may as well be in the next town.

Now why couldn't they do that with all the recordings? I mean was this a mistake. As far as quality goes these two tracks are no different to any others where you cannot do this.

Why don't the record companies who have the master tapes think of re-releasing classic recordings minus the main guitar/bass guitar/keyboard/vocalist or whatever. Would make a lot of money from the amateur musician market, I'm sure.

Those TVS3 mp3s are really impressive and there are 85 of them. Now why don't they offer them for sale I wonder. Those UBH tracks are OK but would rather have the originals minus HBM..who wouldn't.
jay

[Charlie Hall]

Hi jay,
Early stereo was done differently. Usually a lead instrument (maybe with another instrument) panned one side and everything else panned the other side. Later, they could pan everything where they wanted, ie. left, centre, right, or anywhere in between. That's when it starts getting harder to separate instruments.
It wouldn't be possible to derive backing tracks from the masters. Not even from the original session tapes because they only recorded in stereo in those days. Multitracking came a little later.
Regards,
Charlie

[fenderplucker]

Hi Charlie and Jay,

We use a number of different techniques depending upon how the source was recorded.

As Charlie said, if the original had Hank in one channel and the backing in the other, then you can simply replace the Hank channel with your recorded lead.

If Hank was recorded in the center of a stereo mix with the backing spread over left and right, the simplest method is to add the left and right channels with one having the phase inverted. This cancels any audio that was common to both channels (i.e. Hank in the center). This can be done with programs like Adobe Audition (Effects/Amplitude/Channel Mixer) or Steinberg WaveLab. If Hank isn't panned exactly to the center then it will be necessary to vary the mix ratios to get the best effect. This will also remove anything else panned to the center and so it might be necessary to add some back (e.g. bass, strings etc) by recording it yourself or using midi etc. Also, if there was a lot of reverb on the original then this may also leave artifacts.

A more sophisticated way is to restrict the material that is removed, (eg by restricting the bandwidth of the material processed so that bass and cymbals are not affected, for example), by varying the amount of cancellation and also by allowing for some spread across the center channel being processed. This can be again done with programs like Audition, using the Effects/filters/center channel extractor function. The GTX function within JamVOX does a similar thing. Again, it will usually be necessary to vary the parameters to get the best result. However, if you try to go too far the sound will take on an "underwater" sort of character.

Finally, by using spectral analysis it is possible to remove just certain frequency components, either as the main method or to clean up any residual unwanted sounds left over by one of the above methods. This can be done in the spectral analysis view in Audition or more dedicated programs like Magix Audio Cleaning Lab or Sony SpectraLayers Pro. The latter is the best one (though most expensive) as it allows you to select just the fundamental and it will automatically remove all the harmonics. In the others you have to select each harmonic for removal. You might need to also chase down all the echoes as well, so it can be a very time-consuming process. Nevertheless, with this method you can even extract a backing from a mono recording (and this is the only way that one can be recovered from a mono recording, though you will likely go nuts in the process!).

While we have produced a number of such backing tracks, we have done so mainly for our own research in developing the TVS3 to get the closest possible sound to the original recordings (and where it is important to remove just about all traces of Hank's guitar or the result is unfairly enhanced). We respect the ownership of the original material and so have no intention of making them available for sale.

Hope this helps,

Regards,

paul.

[jay]

Great info Paul, that's the sort of thing I was hoping to find out. Now all I need is lots of time, lots of money and a therapist in case I go nuts.....

Will shortly post a list of what I call "vulnerable tracks" for those interested when I've finished going thro' my collection.

jay

[jay]

....here's my list:-

1861
Breeze and I
Cosy
Dance On
Flingel Bunt
Fandango
Nivram
Perfidia
The Rumble
Shazam
Shindig
South of The Border
Tonight
Little Princess
Theme For Young Lovers.

This is for those who are a bit unhappy with the quality of existing full "Shadows backing tracks" and want to have at least something of the originals to play along to whilst practicing at home. I would love it if I could do what Paul has described but I don't think it will happen any time soon.......

Of course, turning the balance control on a stereo system either all the way to the L or R produces variable results....some tunes respond better than others. The Breeze and I is one which is vulnerable and in my opinion is the best example...NP's strings and "cooing vocals" are preserved and when the result is recorded and a CD produced via the computer and played back through the stereo, well at least you get the backing coming through both speakers.

Another thing is that whilst HB is mostly moved "into the shadows" he usually takes one of his bandmates with him, so you get to play with what's left. Sometimes there is just bass and drums and at other times the ryhthm guitar remains but no bass or drums....variable results.

I use various compilation CDs to play the tracks. Curiously, when I played my original 45s of Flingel Bunt and Theme FYL it did not work but when I took the same tunes from a compilation CD it did....maybe Charlie can explain this. I would have thought that whoever produced a later CD of original tracks would have simply transferred them to the new format "as is" but obviously something more is involved?

jay

[Charlie Hall]

Hi jay,
The only thing I can think of is that there are different versions. I am sure I remember someone mentioning that many recordings were done in mono and stereo, and it was a case of making a different recording take for each version.
Regards,
Charlie

[Gerhard]

Hi !

If I have understood right - the goal is to remove the guitar track to achieve the most authentic backing track.

Thinking about another way . . .
Maybe a little bit crazy . . .
There are - as far as I can see around here - so many excellent guitar players which can play every Shad tune note by note exactly with perfect tone. Why not use their skills ?
So what about playing and recording the concerning tune and in a second step mixing it up with the original - inverted phase of course ?
Maybe the procedure has to be done for right and left channel separately.

I'm not an expert - it's just an idea . . .
Perhaps worth to think about it . . .

Cheers
Gerhard

[jay]

Hi Gerhard,

If I understand you correctly........someone records the tune, just the lead guitar part. The recording is then put into software on one track.

Then the original track is put onto a second track.

Then the two are somehow mixed and HBM's guitar is cancelled leaving the new lead guitar pasted onto the originally recorded track.

The result is you, me or anyone else has a recording with HBM replaced.

However, what I would like is to have Hank's guitar removed as much as possible from the original recording so that I can play live
(albeit in my own practice room only) to that recording.

I'm not interested in having my own CD of Shadows tunes with me up front. I might as well just play along with Hank (only louder!)

However, some people may well want to do as you have suggested and I keep hearing about this inversion process to achieve a result.

If it were possible to overdub ,as it were, every Hank note with a "silent note" then I would be interested.....but I really don't know what I'm talking about technically so.......

cheers for now

jay

[Charlie Hall]

It would be great if Gerhard's idea could work, but unfortunately it is impossible to replicate every detail of an original recording in order to then phase cancel it. This was proven when it was suspected that someone who claimed he had played something was found out by comparing his tracks with the originals. The match was too close to the original recording (even though the poster had tried to disguise the recording by re-EQing it to make it sound a bit different) and therefore could not possibly have been played again by someone else.
It is difficult even to phase cancel a part of an original recording with itself! Some artifacts will usually remain, either because the timing is slightly out (even a fraction of a tick can ruin it) or the frequency response is not exactly the same, or artificial or natural reverb cannot be also accurately recreated in order to cancel out the same.
Regards,
Charlie

[jay]

On re-reading Paul's post I see that he has described the process which he used in creating his website's mp3s and if anyone has not heard those then they should go immediately and listen to them because I've never heard anything as good as these, so close are they to the original recorded backing tracks.

Here's the link www.tvsspecialtyproducts.com/images/mp3player.htm

I believe that only some of the 85 tracks were produced in this way with the remainder produced by some other means......

So....what has been done can be done again however difficult that may be for one reason or another.

At the end of his post Paul said:-

"While we have produced a number of such backing tracks, we have done so mainly for our own research in developing the TVS3 to get the closest possible sound to the original recordings (and where it is important to remove just about all traces of Hank's guitar or the result is unfairly enhanced). We respect the ownership of the original material and so have no intention of making them available for sale."

Which I respect.

But this does not mean that no one else can do this.

If I could somehow learn how to do it, then I would. If the end product should not be sold for ehthical reasons,that is one thing.But the know-how behind the product is another and that can be sold or given away for free. Some will agree, some will disagree. Is there any difference between selling the finished product(wrong) and selling the know-how(right)?

If I have got any of this wrong, no doubt Paul will correct the errors and I look forward to any replies.

jay

[Charlie Hall]

Hi Steve,
Nice idea but I daren't allow it for copyright infringement reasons, whether in a members only section or not.
Regards,
Charlie

[Gerhard]

@ Jay

Feb 12, 2013 17:51:50 GMT jay said:

........someone records the tune, just the lead guitar part. The recording is then put into software on one track.
Then the original track is put onto a second track.
Then the two are somehow mixed and HBM's guitar is cancelled leaving the new lead guitar pasted onto the originally recorded track.
The result is you, me or anyone else has a recording with HBM replaced.

No, both guitar tracks, the original and the second one, were "laid over each other (mixed)" but with opposite phase. Imagine two equal sinus waves, equal but the second one's phase inverted. That means when sinus 1 is positive (wave's up) the second is negative (wave's down). When you mix both waves the result is equal Zero - silence. This principle is used for ANR (Actice Noise Reduction) in aircraft headsets for instance.
But unluckyli it doesn't work for our purposes, as Charlie explained.

@ Charlie
Bad news . . .
OK, I had expected some artefact issues, but it really makes no sense when the result is waste. It's a pity . . .
Thought that would be a feather in my cap . . .

Kind regards
Gerhard

[jay]

Gerhard......thanks for the further explanation.
jay

[fenderplucker]

Hi Jay,

I thought I did describe how to do it. If you could let me know what else you need I will try to help. At the end of the day it is really just a matter of getting the software and trying things out.

Regards,

Paul.

[Gerhard]

Hi Paul !

As far as I have understood its a matter of
1. the "right" song or if not
2. the right software, which mostly will become expensive and time consuming.

But when all works fine, in the end you got a BT you only can use in your home for yourself if you want to avoid copyright troubles.
Has that been worth the money and the work ?

Maybe under some circumstances you have to go that way, for development purposes for instance as you had explained.


Kind regards
Gerhard

[jay]

Sorry Paul...... not having much knowledge and even less expertise I did not imagine that I would be able to do what you have done but if you think anyone who is really interested in doing this can do so then I
think I would like to have a go.

Presently, I do not think my PC is powerful enough to do this or is it?

Windows Vista Home Premium
AMD Athlon 64x2 Dual Core 4200+
2.2 GHz
Ram 1.00 GB
32 Bit operating system

And I don't have a DAW either. I tinkered at one time with the JamVox software but wasn't really impressed and I am familiar with what you described as the "underwater sound".......

I was thinking of upgrading anyway. Could I use an iMac with Logic 9?

Jay

[Gerhard]

Hi Jay !

I have no experience with AMDs Athlon Procs, but I have seen you have only 1 GB RAM.
Before you go and buy a new computer I would suggest to give your PC more RAM - Microsoft advises 2GB at least. I would contact the seller of your computer - the optimum would be 4 GB, but it should be the best you can use the built in RAM furthermore. I would advise to ask . . .
On the other hand it depends on the age of the computer - how old is it ?
There are some free DAW programs out there worth to examine, like the Wavosaur Wave Editor (http://www.wavosaur.com/) or / and Audacity (http://audacity.sourceforge.net/).
What you need additionalle is the ASIO4all ASIO-driver (www.asio4all.com/).
I hope I could help

Cheers
Gerhard

[jay]

Gerhard...thanks for that.

I had an idea that 1GB Ram is woefully inadequate. The PC is a Dell
Dimension C521 and is about 4yrs old. I find Windows Vista is frustrating....error messages, things freezing etc. Have had to system restore two or three times this past week just to get rid of unwanted
foreign home pages and search engine pages which have been imposed following internet surfing. Don't know whether the FireFox browser is allowing this? Mac users don't seem to have anywhere near as many problems as Windows users so I have read....

Will have a look at those free DAWs that you mentioned...thanks.

jay

[exprog]

Hi Jay

Your system will take a max of 4GB. If you go to the Crucial site and run the system advisor, it will tell you what memory you can install, and at what price. Nowadays memory is quite cheap, and well worth upgrading.

Chris

[jay]

Thanks Chris....that was v. helpful.

Crucial says I can upgrade to 4GB but a note I read there talks about 32 bit not utilizing more than 3/3.5.

Does that mean that it would be a waste if 4GB were installed or does it mean that 512MB is being used in some other way and therefore it would be OK to go with 4GB?

Maybe I should just upgrade to 3GB?

thanks,
Jay

[exprog]

Windows 32 bit can only address memory up to 3.5 GB. A lot will depend on your cuurent memory configuration. I see you have four slots and they suggest installing in pairs. As you have 1 GB I assume this is two 512MB Dimms.

I would therefore get 2 further 1GB dimms which would then give you 3GB in total.
Any other way would mean getting rid of the memory you have, which is wasteful. 3GB would be ample..

I have an AMD 64 running at 3400, and only have 2GB, which is still OK.

Hope this makes sense..

Chris

[jay]

.....yes it does. I have capacity of 4GB (4slotsx1GB)

Currently using 1GB (2 x 512)+(2 empty).

Think I'll go with your suggestion 3GB (2x512+2x1GB).



Thanks for your help Chris!

Jay