I found a way to have vocals only mixes for early albums

This info should really be in a sticky at the top of the board, but here goes again...

OOPS or out-of-phase stereo is just a small part of a wider form of processing, called phase cancellation. OOPSing is one of the easiest, only works on stereo mixes and only ever produces results in mono, but other much more involved phase cancellation techniques can use mono OR stereo files as source and output. I'll talk about general phase cancelling first, and then describe OOPS at the end, because it's really a limited version of the wider concept.

The idea underpinning all of these techniques is mathematical. If you have x+y and you take away y, you're left with x. Doesn't matter what y IS: that's always going to be true. So if you have a rhythm track comprising bass and drums (call this x), and somebody overdubs ferociously distorted guitar to it (y), you have a track comprising x+y, and you might well not even be able to HEAR the drums and bass any more because the guitar's all over it. However, if you add in 'minus y' to this mix, it cancels out the guitar, and you have the drums and bass again. x plus y plus 'minus y' equals x on its own.

But if you had x on its own to start with, why would you bother to do this? Well, you wouldn't - in this example, you had a recording of the drums and bass on their own already, so no need to go through all this. But with Beach Boys tracks, while we'll sometimes be lucky enough to be in possession of the instrumental backing track of a song (occasionally from a legal release, but more often than not from a bootleg), and the official released version with all the vocal overdubs, we often DON'T have vocals-only mixes (again, over the past few years, an increasing number of these have been released officially, but there are many tracks for which this is not the case).

In theory, by adding a 'negative version' of the instrumental backing (adding 'minus y', if you like) of a song to a released version which consists of the instrumental backing plus the vocal overdubs (y+x), the backing track should cancel out (y+x-y) and leave us with x, just the vocals on their own. These can then be used to overlay on new fan mixes without the backing track getting in the way, or for enjoyment on their own (listening to BW's vocal arrangements in isolation can be a revelation, unmasking all manner of clever harmonic loveliness rather like the backing vocal montage on the new SMiLE set).

What is 'a negative version of the instrumental backing'?. Well, if you've ever seen a display of an analogue waveform in sound editing software, showing changes in volume of a sound over time, you'll know what I mean. A sound can be represented as a series of peaks and troughs, volume changes over time. It looks a bit like a mountain range. If you have a sound, no matter how loud, that has those peaks and troughs arranged in a certain way, and you add an EXACT copy of the same thing, but upside-down, precisely on top of it, the peaks cancel out the troughs, and the result is a flat line: silence. So all you have to do is turn the backing track upside-down, or invert the waveform. This is called inverting the phase of the audio. These days, most decent sound-editing software offers this as a processing option (usually called something like 'Invert Waveform', 'Invert Phase' or 'Reverse Phase').

So: obtain official mix, consisting of instruments and vocals. Easy. Obtain instrumental only mix (much harder). Invert phase of instrumental mix (easy with most any audio editing software), and add to official mix. Result: instruments drop out, leaving you with a beautiful Stack-O-Vocals mix.

That, of course, is the theory. The practice is much fiddlier. In order to completely cancel out, say, a backing track from a finished mix and leave the vocals, you have to have a finished mix and a backing-track-only mix that are sonically identical, and at the same volume. You then have to overlay the inverted version of the backing track ABSOLUTELY PRECISELY on top of the finished mix which contains the backing and the vocals, at exactly the right volume. Fulfilling all of these requirements is next to impossible, and it's much more common that you get a partial cancelling of the backing track, with it reduced in volume but still present to some extent.

To give you an idea of just how accurate you have to be, in a digital recording system running at CD-quality standards (16-bit, 44.1kHz), you have to get your instrumental track at absolutely the right volume, and laid on top of the finished backing plus vocals mix to within one forty-four-thousand one hundredth of a second accuracy. Even with digital audio workstations that work to that level of exactitude, it's still a tall order. And that's just the timing accuracy - the relative volumes of the two tracks have to be exactly right too.

On top of those already tough difficulties, there's the issue of where you might get an exact instrumental version of the track you want to process in this way. Broadly speaking, the more unlike the finished mix the instruments-only mix sounds, the less well the instruments will cancel. So if you got an instrumental mix from a bootleg MP3 file, the MP3 compression process will have rendered the backing track sonically slightly different from the one you're trying to cancel out of the finished mix, and it won't fully work. If the bootleg came from a dodgy ninth-generation cassette tape of a working mix reel that somebody nicked from the Beach Boys archives in the 1970s (say), the speed variations in the cassette copy (and the swathes of hiss that the cassette will have added) will mean that the backing will not sit properly on top of the finished mix, or be close enough sonically to the finished mix, to cancel the backing out perfectly. If the bootleg backing track is mono, and the track you're trying to cancel has the backing on it in stereo, that difference will definitely mean it doesn't work, either. If you have an instrumental backing track that was mixed recently and has had reverb added to it to make it sound more palatable on its own without the vocals, and you are trying to use that to cancel from a mix made in the 60s without the reverb and on much older equipment, the chances are that THOSE sonic differences will make the backing track you have sufficiently different, sonically speaking from the backing track in the mix, that it won't cancel out then EITHER.

In practice, unless you have a copy of an instrumental mix made at the same time as the full mix with the vocals that you're hoping to process, and unless both mixes were done on similar equipment and featuring a similar overall sound balance and mix, full cancellation is next to impossible. Occasionally, though, we do have these kinds of mixes for Beach Boys stuff on bootlegs like the Sea Of Tunes CDs, where you'll hear a backing track mix, then hear a very similar mix with overlaid vocals, and because the two were done at very similar times and are sonically very close, you can sometimes cancel the backing and extract the vocals. Even then, speed variations between the two mixes can often mean that the cancelling works for a short time, and then the synchronisation slips and the inverted backing-only mix stops cancelling the official mix. So you hear the vocals isolated for a time, and then the backing fades back in underneath, and very soon, it's not cancelling at all any more.

That's full phase-cancelling, and if you are lucky enough to have the right audio bits to start with, accurate digital editing software, and a lot of patience with synchronising waveforms, it can yield some amazing results. Although I've talked about cancelling instrumental backing tracks to obtain vocals here so far, that's just an example, and it doesn't have to be used for that. Going back to the maths again, if you have a mix comprising x+y, and you have a decent copy of y, you can get x in this way - but x could be anything. If you have a recording of a finished mix and you have the vocals, you can get the instrumental backing. If you have a recording of an orchestra and also the exact same recording but with a guitar overdubbed on it, you can get the guitar out. If you have a finished vocal harmony section and the same section with a lead vocal overdubbed, you can get the lead vocal. To take an example from the new SMiLE Sessions set, if you have the Vocal Harmony insert from Vega-Tables with the overlaid lead vocal, and you have a mix with the backing harmonies only, you can get the isolated Brian Wilson vocal. And in theory (this is kind of the Holy Grail of all Beach Boys-related achievements in this regard), if you have uncompressed version of the backing track to Good Vibrations (as released in 2006 on official CD single and now again on the SMiLE Sessions box) and you invert it and overlay it on the uncompressed 45 single mix from 1966, the accapella vocals will drop out.

However, that's the theoretical maximum you can achieve, again... whereas more often than not, the whole process is scuppered by you NOT having quite the right material to work on to begin with. When I first started experimenting with this technique, I spent many hours trying to cancel a stereo backing track using a mono copy of the backing track. I spent a very frustrating time wondering why it wasn't working, hours of sliding waveforms over each other and adjusting their relative volumes, looking for the perfect sync point where they would precisely overlap and match in loudness so that they cancelled out, before I eventually realised that I had completely wasted my time. The two mixes, although they were of the same song, were just mixed too differently for one to be able to cancel the other. And in practice, this is what messes up the chances of getting a decent copy of the GV vocals. The 2006 instrumental mix is in stereo and was edited together and mixed in 2006. The 45 single was edited and mixed in mono in 1966. They may be the same song, but the mixes are different enough that a complete cancelling is quite difficult. It can be done, but bits of the backing track remain audible in what you end up with, and vary in volume throughout. And simply rendering the 2006 backing mix INTO mono before you start doesn't fix everything either. The mix is STILL too different, stereo or mono.

So that's 'true' phase cancelling, with an idea of its relative merits and disadvantages. OOPS, on the other hand, is a simultaneously more limited but much, much easier to use phase-cancellation technique. With 'true' phase cancelling, you can use stereo or mono files as your start points. If you have a stereo mix of a song's backing and the same recording with 17 blistering overdubbed electric guitars going at it in stereo, you can, in theory, cancel the backing to obtain a stereo mix of the guitar overdubs. Similarly, if you have a mono brass band recording and the SAME mono recording with a choir overdubbed in mono or stereo, you can in theory cancel out the mono brass band and obtain the mono or stereo choir. Provided you have the right parts to work with (and as we've seen, that 'provided' is one of the tricky bits...), the format of the input or output is not an issue.

With OOPS, you can only use a stereo file to start with, and the result is always mono. Essentially, instead of overlaying completely separate audio recordings on top of each other, you invert one channel of the stereo mix of a song (say the left channel), and overlay it on the other one (say the right). As with 'true' phase cancelling, what is common to both channels is instantly eliminated, and you're only left with what was different in each channel. The upshot of this is that OOPSing cancels out whatever is in the dead centre of a stereo mix, removing it completely, and leaves you with what was mixed differently between the two channels. This is why OOPSing mono doesn't work, because in a mono mix played on a stereo system, both channels are the same. In that case, when you invert one and add it to the other, the cancellation is total, resulting in silence. It's also why the output of OOPS is always mono, because in adding the two stereo channels together to cancel what is common to them both, you create a mono track.

If you can live with that, though, the advantages are that you don't have to spend hours trying to precisely overlay the waveforms on top of each other, or adjust their relative volumes much, as they're already locked together and balanced in the stereo mix. But the relative usefulness of the technique can be much less - it all depends on how the original track is mixed.

In a lot of modern music, lead vocals *tend* to be panned to the centre of stereo mixes, which means that these vocals often drop out when you OOPS. OOPS is often known as 'vocal cancelling' as a result, and OOPS techniques are often packaged as 'vocal cancellers' or 'vocal removers'. However, these are misnomers, because all OOPS does is remove whatever is in the dead centre of a stereo mix. If the vocals haven't been placed there, they won't be cancelled or removed. If the lead vocal has stereo reverb applied that makes it unequally distributed in the left and right channels, the vocal may not be completely removed anyway. And OOPS is certainly no good for removing stereo backing vocals, which are often (particularly in Beach Boys stereo mixes) panned left and right, well away from the centre of the stereo soundstage. Finally, OOPS certainly doesn't ONLY remove vocals, even if these are panned dead centre - it removes whatever is panned there, including bass and drums, which are often placed there in modern mixes, so you frequently end up with very strange audio that has no vocals, no bass, and no drums, but may still feature keyboards, percussion, and backing vocals, all mixed together in a messy mono mash and not much use to anyone. OOPS is also utterly useless on those strange unbalanced mixes from the early days of stereo, like some of the Beatles mixes from the late 60s. Often very little was placed in the centre of these, vocals could be on the left and drums and basses right, and in these cases, OOPSing will do very little or nothing at all.

For all that the usefulness of OOPS techniques is subject to the vagaries of how the original track was mixed, it turns out that it IS unexpectedly useful on many Beach Boys recordings,
particularly the odd (by modern standards, anyway) stereo mixes created by Chuck Britz for the early Beach Boys albums. As explained, when these albums were recorded, it was still early days for stereo, and Brian only cared about the mono mixes anyway, so the stereo mixes seem to have been something of an afterthought. Many of the tracks were mixed with the backing tracks dead centre, and the Beach Boys vocals panned left and right. Few mix engineers these days would mix tracks in this way, but with the earliest albums, there may not have been much freedom anyway, as many of them were recorded on three-track machines at best, which might only allow a mono track for the instrumental backing, and two other tracks, one for group vocals and the other for a doubled pass of the vocals, or occasionally the overdubbing of another instrument. If that's all you have on the tape, how else *could* you mix other than backing centre, and vocals left and right?

However, to those interested in OOPS techniques, these oddball mixes are gold. Far from partially or fully cancelling a bass guitar, or a lead vocal, as OOPS techniques usually do to modern mixes, when these early 60s 'backing centre, vocals left and right mixes' are put through the 'OOPSer' (so to speak), the BACKING TRACKS drop out, leaving the isolated vocals (albeit in mono). But it's just an accident resulting from the specific way in which those early albums were mixed, and it certainly won't work on everything. From the ...Today album in 1965 through to Friends in 1968, everything was only mixed to mono back in the 60s, so OOPS cannot be used on the official release mixes of Today, Summer Days, Party, Pet Sounds, Smiley Smile and Wild Honey. (And before anyone chimes in, yes, of course artificial 'Duophonic' stereo versions of these albums were made, but OOPS doesn't work on those in any useful way.)

Interestingly, and whether deliberately or accidentally (it may have been either), many of the tracks on the Sea Of Tunes bootlegs series (though not all) were also mixed in a way that makes them usefully susceptible to OOPS, such that the vocal and instrumental overdubs can often be isolated from the mixes by careful use of the technique. What often happens on SOT is that an instrumental or vocal overdub will be featured in a mix panned left or right, while the backing track to which the overdub was performed is presented in the centre. In these cases, OOPSing removes the backing and leaves the isolated overdub. Instrumental and vocal parts for many of the albums that were only ever properly mixed to mono in the 60s (Today to Wild Honey, and also much of the SMiLE material presented on SOT Volume 17) can often be isolated in this way.

OOPSing is not something that can only be done in software. It used to be done electronically in the days before digital editing software, you can do it by re-wiring connection plugs between a mixer and a recorder, and you can even arrange for it to happen mechanically, by half pulling out headphone plugs, as AGD mentioned in an earlier post in this thread. And, as punkinhead said at the start of this thread, it can even happen by accident when your plugs go bad (which explains why he suddenly found himself listening to mono accapella mixes of the Beach Boys singing in his car - it's not that 'the backing track channel was turned off', or that 'only the vocal channels were left on', as there aren't separate 'vocal' and 'instrumental' channels for those albums; it's just that those albums are mixed in a certain way, and the inadvertant OOPSing on those tracks resulting from the duff wiring in his plugs happened to cancel out the centre portion of the stereo mix where the backing tracks happened to be mixed on that album, leaving just the vocals).

While you *can* make OOPS happen in all of the above ways, it's far more controllable, useful and rewarding in software. Best of all is when you have control of the volume of one of the tracks from the stereo mix on a fader and can raise and lower it while the volume of the other (phase-inverted) channel remains at one hundred percent. When the fader you're controlling is at full volume, everything in the centre of the mix cancels out. As you slide the fader down, you can sometimes (again, depending on how the original track was mixed) achieve cancelling of other parts in the mix placed at different points in the stereo soundstage, moving progressively towards one side of the mix or the other - although of course, as you move the fader down, whatever was mixed in the centre is *uncancelled* and you start to hear it again. Nonetheless, such a setup can be a very versatile way to listen 'through' and 'into' the parts that make up the mix, and can give you a much better understanding of the production techniques used on the track, if you know what you're listening to.

To finally answer hypehat's question, there will probably be some cool stuff that can be achieved through OOPSing and also pure phase cancelling with the SMiLE Sessions set. However, it's not going to be totally straightforward, and without instrumental-only mixes of most of the tracks, it certainly won't be possible to just obtain lead vocals just like that. For one thing, if it was just isolated lead vocals you were after, you'd need to have mixes to hand to cancel with that featured everything EXCEPT the lead vocal: instrumental backing *and* backing vocals, mixed exactly the same as the main album tracks but lacking the lead. As far as I know, there aren't any tracks on the set like that.

What's more, OOPSing will of course be useless on the Disc 1 'assembled SMiLE', as it's all in mono, and phase cancelling stuff from the mixes on CD 1 will be difficult using the material on the remaining discs because the CD 1 stuff is mono and the rest is stereo. And 'folding down' the material on Discs 2 to 5 to mono changes the mix in such a way that it no longer overlays exactly on the Disc 1 mono mixes, or cancels reliably from them. Again, it all depends on how the material is mixed. But nonetheless, I think some interesting stuff will be possible. It should be possible to cancel 'Truck Driving Man' out of Cabin Essence, for example, because what you have there are two sections mixed the same, but one is different by dint of having the extra overdub for Dennis's vocal. Subtract one from the other, and you should be left with Dennis's vocal on its own. But more than that, I can't say without checking out some of the mixes in more forensic detail. I will report back, at some point...

MattB

The Smiley Smile Message Board