Re: sampling

Creative Sampling: Part 2

Published in SOS January 1999
Printer-friendly version

Technique : Sampling

Even the more basic sound-manipulation tools offered by most samplers can lead a safe sample into a world of weirdness. Nicholas Rowland reverses, stretches, bends and loops, all in the interests of sampling science...

Since last month's article, which investigated potential sound sources for samples, you'll hopefully have all been out with your Dictaphones throwing muses down wells, around squash courts and against metal garage doors, and recording the results. This month we look at a few ideas for what to do with sounds once you get them home and in front of a blank sampler LCD. So without further ado...

Usually, the first stop for creativity in sampling is to hit the reverse key. After all, it's easy enough to do and scores 11 out of 10 in the instant gratification stakes. Aside from turning the most inane chatter into Exorcist-style devil talk, it's useful for turning sounds with sharp attacks and long decays into accents and fills for that essential dance mix. You can play it with more subtlety, though. For example:

• Try reversing a looped sound (percussive or ambient) and mix this in with the original. Or try recording a sample with tempo delay and then reverse it. It sounds more eerie than you might expect.
• With stereo samples of ambient or melodic sounds, try reversing one channel for a more unusual stereo image. You can also play around with panning here, perhaps crossfading each one over to the other side of the stereo spectrum as they play.
• You can create the illusion that certain sounds have been reversed by coupling a very slow attack time with fast sustain and release times. Try running a backwards-sounding version of a sample against a true reversed version of the same sample as the two halves of a stereo pair, panned hard left and right.
• Some samplers (the Roland W30, for one) have a function which makes the sample automatically play forwards and then go into reverse. It's designed as an easy way of elongating sounds without faffing about trying to find loop points, but with rhythmic loops and percussive sounds it can be a great way to create some unusual effects. All I can say is, if you've got it, flaunt it. If you haven't:
• Copy your original sample, reverse it and then stitch it on to the back end.
• Try panning the sample across the stereo field as it plays, with the centre point being the point at which it starts to reverse.
• Drop the reversed element down a semitone or two to create a psuedo-Döppler effect.

• Simple flange and phasing effects can be achieved by double- or triple-triggering the same sample.
• For a chorus effect, do the same and detune the second and/or third sample by a tiny amount.
• A useful trick involving modulation, which works well for pads and string sounds, is to make copies of the same sample and then apply different vibrato rates to each one to create a chorus effect.
• Interested in crunchy loops?
• Try sampling at the lowest bandwidth your sampler offers.
• Try deliberately sampling at too low a level, then using the normalising function repeatedly to pump the volume back up again. You'll find your pristine sample soon gets mucked up with all kinds of digital silage.
• Give rhythm loops extra crunch by over-normalising them so that the peaks get clipped.

Another grungelising trick is simply to record the sample to tape, then record it back into the sampler. Keep doing this until you're grunged out. The noise reduction on tape players can also add to the character here -- try recording back to tape with the noise reduction on, then record to the sampler with the noise reduction off. If you're lucky you'll get a pumping effect which can often only add to the charm of your original loop.

If you loop a phrase on your sampler and then run it alongside a sequencer, you usually find that, unless you're 100% accurate in setting the start and end points, the loop will start to drift out of sync with the rest of the track. This can be irritating if you're trying to achieve a tight rhythmic effect, which is why most people will retrigger the sample from the sequencer at the start of every measure.

However, with more atmospheric loops it can actually be quite effective to have the sample gradually (or even dramatically) drifting out of time with the main rhythm. You could even try deliberately running two versions of the same loop together -- one triggered regularly on the bar, the other trimmed to fall, say, a 32nd-note short of the full bar. If my maths serves me correctly, this means that they will be in sync only once every 32 bars... spooky! Naturally, unless you're deliberately trying to psych out your listeners, this technique is probably best confined to background ambient sounds rather than the more in-your-face drum loops. Pan the two samples to opposite sides of the stereo spectrum for added eerie points.

As all good samplists know, a sample doubles its speed when it's transposed up an octave. So try triggering two versions of a sampled loop an octave apart. With a percussive loop you'll get instant plinky-plonky percussion running over the top of your original. With melodic loops, the results are more unpredictable, but by playing around with the intervals, you can get come up with some interesting arpeggio-style sequences.

Incidentally, when trimming samples that are going to be used as rhythmic loops, you can give yourself a helping hand by pitching the sample down one or even two octaves. Slowing the loop down makes finding the 'right' start and end points much easier. When you've finished, put the loop back up to the correct tempo/pitch and away you go.

Once confined only to top-end samplers, time-stretching -- changing the length of a sample without affecting its original pitch -- has now become a standard feature on virtually all current models. It's an essential tool in dance remixing and live DJing work to fit rhythm loops to the tempo of another track, or to re-pitch vocal phrases without getting the old chipmunk effect.

Basically the sampler chops up the original sample into smaller chunks, then strings these out across the new time frame, doing its best to fill in the resulting gaps by extrapolating information from the bits of sample either side. The more you time-stretch a sample, the bigger these gaps are going to be and the more thinking the sampler has to do to fill them up in an intelligent way. Stretch a sound too much and you get all kinds of weird results as the sampler struggles to make sense of it all. While this is not so good if you're going for hyper-realism, it can be seen as a route to mangling heaven. If in doubt, experiment. Even if you want the sound at its original pitch, try time-stretching it, record it to tape, then resample it and shrink it back to the initial length. What gets lost in the translation may just work to your advantage.

An alternative to time-stretching an entire rhythm loop is to use the pitch-bend wheel to stretch parts of it just enough to make it fit the new tempo. Of course, you'll experience a change in pitch of the sample somewhere along the line, but this can add to the charm of the rhythmic experience. It takes a bit of trial and error to get right, but if you record the movement of the pitch wheel as part of a MIDI sequence, you'll be able to fine-tune it so that the bent rhythm fits the new tempo exactly.

For those without time-stretch functions, but with infinite patience and a calculator, there's an alternative DIY method:

• Imagine you're starting with a four-on-the-floor, 16th-note loop that's been first recorded and edited precisely to length. You'll first need to find out the length of the loop in actual samples (or if your sampler works in milliseconds, you'll have to work in those).
• Now divide this number by the appropriate value. If it's a 16-beat rhythm you'll need to divide it by 16; if it's eighth-note triplets you'll need to divide by 24.
• The resulting number gives you the length of each beat and from this you should be able to work out the relevant start and end points. So let's say your beats are 5000 samples long: the first beat will run from 0-5000, the second from 5001 to 10,000, and so on. You can now use these figures to cut the rhythm into its 16 component beats. In some cases, you'll have to keep on copying the sample to trim each bit, and with samplers and hard disk recording systems that employ non-destructive editing, you'll be able to cut up your original loop 'virtually'. Note that you may also have to fiddle around with the start and end points of each individual beat, so that parts of one beat don't spill over into others.
• With careful editing you'll end up with 16 different components which you can then assign to individual MIDI note numbers so they can be played as individual sounds. (Make sure you're triggering them at the original pitch.)
• To play the loop from your sequencer, you will then need to set up a run of 16th notes to trigger the beats consecutively. At this point you can start playing around with the sequencer tempo to either lengthen or shorten the loop.

If you've done it right, running the loop faster doesn't usually present too many problems. At tempos which are considerably slower than the original, you will find that gaps start appearing. This can be masked to a certain extent by playing around with the attack and decay of the samples to get rid of any abrupt gaps and then adding a touch of reverb to mask the gaps. Or you can lengthen the sample by tacking a reversed version on to the end of it and then shaping it with the sampler's ADSR envelope.

Of course, you can get a bit more creative here, applying different parameters to shape the different beats in various ways. And as you've now broken your original loop into its component parts, you can also trigger it in a different order, or isolate various elements and use them to create fills and breaks. Suddenly, what started as an exercise in DIY time-stretching becomes a brilliant way of creating new grooves.

If all this sounds interesting, but too much like hard work, you can always cheat by going to the shops and buying Steinberg's Recycle for your PC or Mac.

Any sampler worth its salt will respond to MIDI velocity. Samplers with the full condiments will also allow you to assign other parameters to MIDI velocity (and, indeed, any other continuous controllers as well). To make samples sound more dynamic:

Thanks For The Memory     Limited sample memory is not such a big issue these days, as memory is cheap and many samplers have plenty of room for expansion. However, Murphy's Sampling Law states that however much you've got, you'll always need more. So here are some tips on how to save it. Use stereo only if you need it, as it takes up twice the memory of a mono sample. Admittedly it can make for more interesting sounds, but for drum loops, basses and vocals you're often better off sampling in mono and then adding pseudo-stereo through external effects. If in doubt, cut it out. Get into the habit of trimming samples, particularly the back end, to get rid of any dead space. Be frequency-conscious. If your sample doesn't have much top end -- or you're looking for an interesting lo-fi sound anyway -- sample at lower bandwidths wherever possible.

• Try assigning the filter cutoff point to velocity, so that when you play harder, the sound will not just get louder, it will also get brighter.
• This works well with 'real' drum sounds, which get brighter and have more attack the harder they are struck. If your sampler allows, assign a slight pitch offset (say 10-20 cents) to simulate how acoustic drums sound slightly higher in pitch when they are hit hard. Guitar sounds can also benefit from this approach, though you have to make sure that the pitch bend isn't too extreme, or it will just sound as though your virtual guitar has gone out of tune.
• You can also try this trick with other parameters, such as pan position -- this way, a staccato snare fill will also move across the stereo field as it gets louder. Again, if your sampler has internal effects, you may be able to link velocity to the reverb or chorus send, so the sound becomes more (or less) effected at higher volumes.
• If you're working with more rudimentary equipment, you can achieve similar results by layering samples and then crossfading between them. (Incidentally, if your sampler doesn't support crossfading as such, simply assign two or more samples to the same key group, then play around with their respective velocity curves to fade one in as the other fades out.) For example, try recording a sound dry, then recording a second version with more attack, extra resonance, a touch of reverb or chorus, or a combination of all of these. Then set up your sampler so that the effected sound only begins to fade in at higher volumes.

Not only will your samples sound more interesting, the fact that their character changes according to velocity means that they will be more inspiring to play. This technique works well with organ and guitar sounds when you have a 'straight' sample combined with a distorted version that kicks in as you play harder. Samplers that respond to aftertouch will also enable you to add a third level of transformation which, again, can be used to add dynamism to sounds.

When setting up a sample that's going to be used across the whole keyboard, you can reverse-engineer this process. As the sound gets higher up the keyboard, give it a sharper attack, a significantly longer decay and more of a pitch offset.

The most creative implements in the samplist's toolbox are the ones stuck on the sides of your head. If it sounds right, it is right. I find that the most useful samples come from my own recordings -- not because I'm completely brilliant, you understand, but because the samples are already of the right kind of character to fit the music. So although it's not a sample manipulation technique as such, get used to listening for those golden moments in your own material. Sampling extracts from your own music, and running them through EQ and effects, could send you off on an unexpected journey into sampled sound...