Sampling in Stereo
Your sampler may only have a mono input, but that shouldn't stop you taking two-channel samples for a fatter, more realistic sound. Howard Massey shows you how.
Your sampling keyboard may not have the stereo inputs, but using some simple techniques, you can make your own stereo samples - and reap some rich rewards.
THE AMOUNT OF available memory in these instruments is largely a function of their price, though as raw computer memory gets cheaper, we shall undoubtedly start to see more and more powerful samplers available for less and less cost.
Hugely expensive samplers like the Fairlight Series III and Synclavier not only offer higher-fidelity samples, but allow you to sample literally minutes (if not hours) of signal. Because their hefty price-tag implies mammoth amounts of onboard memory, you can not only sample monophonically in these instruments, but you can actually feed a stereo signal in and sample in stereo - though this quite naturally halves the total amount of sampling time available to you.
Smaller, more affordable samplers like the Emulator II, Emax, Prophet 2000, Akai S900, Korg DSS1, and Ensoniq Mirage have much less available onboard memory (generally well less than a minute of sampling at any kind of decent fidelity), and so don't offer such an option.
But recording engineers everywhere know the value of working in stereo. Recording two simultaneous audio signals instead of one makes for an audio experience that is far closer to "reality" - after all, those of us not named Van Gogh have two ears, not one. Synthesisers with stereo outputs generally produce sounds of a richer quality, largely because of the inherent phase cancellations and reinforcements caused by stereo spatiality.
Furthermore, virtually all final recordings these days are stereophonic. Samplers are today increasingly becoming an extension of the audio tape recorder, as they allow manipulation processes like looping, splicing, and editing to be performed easily and without the need for barbaric devices like razor blades. (Shave and a haircut, please, and while you're at it, can you edit in a new chorus...?) One of the most common uses for sampling in a recording situation is to feed a word or phrase directly from the multitrack recorder into the sampler and to then "break, break, brbrbr breakbreak-break" to your heart's content, in real time to the final mix. More often than not, however, vocals (particularly backing vocals) as well as instrumental tracks have been submixed into a stereo image. If you then take that stereo image and sample it monophonically, it's obviously going to take a fair bit of tweaking to then lay it back into the stereo mix with any kind of continuity.
"Recording two simultaneous audio signals instead of one makes for an audio experience far closer to 'reality' - after all, those of us not named Van Gogh have two ears, not one."
SO WHAT DO you do? Hire in a Fairlight? Fine, if you've got the dough. But if you already own one of those less expensive samplers, you can pocket your money, because what I'm going to do here is tell you how to obtain stereo samples and play them from your existing machine.
This technique was perfected on an Emulator II, which is the sampler I find myself using most often, but there's no reason why it won't work on many of the other aforementioned samplers just as well. The only real conditions are that: 1) your sampler must offer you the ability to assign two voices to a single key; 2) it must provide separate outputs for each voice (this leaves out the Mirage, I know - sorry about that); and 3) it must give you the ability to truncate the beginning of the sample.
It will also be useful, though not mandatory, to have an adjustable audio threshold setting in order to actually start the sampling process.
First of all, of course, you'll need a stereo source signal. This could be a record, tape, or CD, or it could be a stereo output from a single stereo synth, or it could be a stereo submix from either a tape or a MIDI system. Let's presume, for this example, that our source signal will be the stereo output from a submix of several backing vocal tracks taken from a multitrack master.
"We need to synchronise the two sides so that the total stereo image begins at the same time, even if one channel starts a bit later than the other. Let's use good old-fashioned physical reality - tape."
One problem that needs to be dealt with is that the signal in the left channel may not begin at precisely the same time as that in the right channel - after all, in a stereo submix you can place any part of your total sound anywhere in an imaginary 180-degree plane. Let's suppose, for example, that the backing vocals are singing the phrase "stick it in your ear" and that the male backing singers are singing right at the start while the female vocalists just come in for the "in your ear" part. You might want, for example, the male vocals in the right channel and the female vocals in the left. Or you might put all the vocals on one side and a short digitally delayed version on the other. You see what I'm getting at, I hope - the point is, both channels won't necessarily start at the same time.
So we'll need to somehow synchronise the two sides so that the total stereo image begins at the same time, even if one channel starts a bit later than the other. Does this mean we need to resort to an arsenal of expensive sync boxes? No way. Let's get more basic than that, and use good old-fashioned physical reality instead. Meaning - tape. Or cassette. Or (best of all) a VCR tape medium for digital recording (like the Sony PCM systems). If we dub our stereo source signal onto one of these media and precede it slightly with some kind of recognisable percussive signal, then clearly, the distance between the start of that percussive signal and the start of our stereo source material (the stuff we really want to sample) will remain constant, so long as the tape or VCR speed remains constant.
Therefore, do the following. Record some kind of sharp, percussive sound onto your tape, cassette, or VCR (if you're using the PCM system), making sure that it is recorded on both tracks at the same VU level. A second or so later, record the actual stereo sound you want to sample. Try to keep the gap between the two as short as possible, for reasons that will be obvious in just a moment.
The real purpose of this percussive "leader" is simply to trigger our sampler into starting the sampling process - assuming you're working with an instrument that provides you with an audio threshold control. If your particular sampler doesn't have that feature, however, there's a way around that problem, though one which isn't quite as precise. Instead of recording a single percussive sound, record four of them in rhythm to act as a kind of count-in. What you'll then need to do is manually start your sampler sampling on the fourth beat. If you've got any kind of decent sense of rhythm, this should work pretty well.
"Having taken great pains to ensure that both samples started at precisely the same instant, we can truncate them at the same starting point with confidence that they will remain synchronised."
Obviously, since none of the samplers we're discussing here has a stereo input, we'll have to sample each channel separately. If your instrument has an adjustable threshold control, be careful to keep it at exactly the same level for both samples. If you're not, try to start the sampler precisely on the fourth beat both times. If you feel under-qualified to do this accurately, call up your sister's boyfriend who happens to be a drummer and have him hit the button for you. The trick, in either instance, is to get the start of the sample the same for both channels. This will ensure synchronisation of the final signal.
The next step, after you've sampled both the left and right channels (and saved them to disk - you don't want anyone tripping over the power cable after all this hard work) is to edit out the percussive trigger. Listen to both samples individually, and determine by ear which one seems to start earlier - this is the one to edit first. If they both seem to start at the same time relative to the percussive trigger, then you can work with either one first. While monitoring just the one sample you've decided to edit first, use your instrument's truncate feature to remove everything from the percussive trigger, right up to the start of the sound itself. Before you make the truncation permanent, write down the value of the new start point (in bytes or seconds, whichever way your instrument displays it). Now go to the other sample, and without even listening to it, simply truncate its start at exactly the same point.
Since we took great pains to ensure that both samples started at precisely the same instant (through either the threshold control or your sister's boyfriend's phenomenal sense of rhythm), we can truncate them both at the same starting point, with confidence that they will remain synchronised.
All you need do now is assign both samples to the same key or series of keys (up to a two-octave range on the Emulator II, possibly more on other instruments), and then utilise the individual outputs offered by your instrument. Feed the sends into your mixing desk, and pan them hard left and hard right accordingly. On the Emulator II, you can achieve as much as four-voice polyphony of your stereo sample by assigning the left sample to channels 1-4, and the right sample to channels 5-8 - then simply pan channels 1-4 of your mixing desk to hard left and channels 5-8 to hard right.
THE SUBTLE BUT important phase differences in a stereo image should immediately yield huge benefits when you listen back to your stereo sample - you'll undoubtedly find that your sound is immediately "bigger" and far less directional, which will in turn make it much easier to mix in with your final stereo signal. We've gone a long way since the days of mono, after all.
There's no reason why your sampling instrument should be providing only monophonic sound sources if it has the capabilities outlined above, most of which are pretty standard these days, anyway.
Try it. You'll like it.
Feature by Howard Massey
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!