Shaping Sound (Part 3)
Filtering, Enveloping And Keyboard Sensing.
In the penultimate part of his series, designed to help you get the best from your sampling machines, keyboard programmer Paul Wiffen returns to his analogue roots to modify some samples.
Continuing his quest to combat uniformity, keyboard programmer Paul Wiffen reveals how you can apply traditional analogue sound-shaping techniques to improve your sampled sounds.
We ended last month's session with our sound sample chopped down to size, possibly spliced or mixed with another sound and generally taken about as far as the digital process currently allows. To go any further we have to resort to old and much-loved (at least by me) techniques, such as filtering, enveloping and using velocity control. In other words, back to our analogue roots.
The reason I am so sceptical about expensive all-digital systems such as the Fairlight, Kurzweil and Synclavier, or even the more affordable Yamaha FM gear, is that, however great the sounds you produce, the manipulation techniques offered by the machines are complicated and fiddly - a bit like taking a nutcracker to smash a paving stone (the opposite of the old sledgehammer to walnut syndrome). For all its benefits, digital technology has yet to come up with anything as simple, quick and effective as the good old VCF (voltage-controlled filter - or more likely digitally controlled filter these days). Casio's Phase Distortion is the quickest technique yet to come from digital sources, but even this hasn't the excitement and sheer butchness of an analogue filter in full sweep.
It is for this reason that I actively prefer to work with digital/analogue hybrids like the Prophet 2000, Emulator II, PPG or even Ensoniq Mirage (in order of preference) against the most expensive all-digital leviathans, even if the budget available could cope with the outermost excesses of New England Digital. It was with great relief that I learned direct from the horse's mouth that the Series III Fairlight will finally make dynamic filtering available on its superior fidelity sampling. Sounds like they've seen the light at last.
Anyway, enough of this bitching, and on with what this wonderful device called a Filter can do for your samples. Now, those of you who know your analogue onions, will be able to name three standard filter types which appear on synthesizers. The ubiquitous Low Pass (which every analogue synth features as standard) and the rarer Band Pass and High Pass (which only appear as options on more recherché machines like the OSCar, Korg MS range and Elka's Synthex). While these last two may be useful in more advanced synthesis applications, all they will achieve is to highlight the problems we spoke of in the first article, namely, poor signal-to-noise in the higher frequencies, and aliasing effects.
What we need here is something which will take these problems away and here we resort to the trusty old Low Pass Filter. As this lets the fundamental through and attenuates the higher frequencies, this is just what we need. Of course, we must be very careful in our use of it or we may find we end up with sounds that appear as though they were sampled on a 2-bit system!
If we use the filter statically (as you can on the Fairlight, for example), then we will seriously reduce the brightness of the sound. In this application you can only really get rid of serious aliasing effects (for example, when the sample is transposed down more than an octave), or high frequency buzzes or other frequencies which have been inadvertently sampled. Of course, this does allow us to overcome really drastic problems with a sample and so is a useful addition to our arsenal.
However, where the Low Pass Filter really comes into its own is when it is dynamic (ie. it moves during the course of the sound). This used to be done by means of voltage control generated from an ADSR envelope, but these days the envelope is more likely to be of digital origin. Still, analogue filtering is exactly the same. What this means is that you can have the filter fully open for the start of the sound (a fast attack), where nine times out of ten the sample is loudest (pianos, guitars, percussion and almost any instrument blown or bowed hard), and then use the decay, sustain and release portions to follow the natural dulling which occurs when sounds die away, filtering out noise and any other undesirables, but leaving the lower frequencies which hang on longer untouched (see Figures 1a, 1b and 1c).
This technique works particularly well on somewhat noisy sampling systems like the Ensoniq Mirage (which generates a fair amount of noise in addition to that present in the original sound source). If you look at the envelope settings of the Mirage's factory sounds, you will see the values (approximately) occurring again and again. It is by such crafty means that the outstanding Library Disks from Ensoniq are produced.
Of course, you don't have to use envelope shaping in this way. Once again I would make a plea for creativity. Don't let imitation be your sole aim. Try improving upon natural sounds and use your judgement to decide what is a good effect. A slowly opening filter (high Attack number) can completely change the character of a piano or string sound, whilst a quick Decay turns it into tuned percussion or pizzicato. Nothing is wrong - it either works in a certain piece of music or it doesn't.
Amplifier enveloping is a common feature of virtually all samplers (except those on certain drum machines) in that when the keyboard is released, the sound is cut off. This is usually instant unless the Release rate is increased. But again, tidying up of sounds or alterations can be achieved with VCA-type ADSRing. Be careful here though. You are not just attenuating higher frequencies now, but the entire signal. If that's what you want, all well and good, but be sure you don't lose the fundamental too soon, especially when you play the sample in the context of the piece of music.
A slow amplifier attack can give you the same effect as what guitarists call 'bowing' (using the little finger to turn up the volume knob after the string has been hit). Try this 'bowing' effect on piano samples or any other sustained sound that has a sharp attack. On short duration sounds, the loudness envelope isn't so useful for a slow attack just quietens the sound or loses it altogether.
Generally speaking, you will find that Filter and Amp Enveloping is most effective on sounds which have been sampled long or looped. This gives you sufficient time to manipulate them. However, it is possible to make use of filtering and loudness in conjunction with a more recent innovation in keyboards: velocity control.
When an instrument is played hard (be it piano or drum, violin or trumpet), there are two effects: the sound gets louder and brighter. Soft strikes tend to be not only quieter but duller, so we can press our Filter and loudness control into service here too. (See Figures 2a/2b.)
Take a sound which has been sampled at the hardest play you will need. Set both Filter and Amp to fully open on the envelope side and then increase the velocity amount to both. This will make softer key strikes close the Filter and attenuate the output. Obviously, you will need to alter the amount of each for different samples, as some sounds vary a lot in volume but not in timbre, whilst for others the converse is true. Practice listening to instruments and hearing the loudness and brightness changes that result from different key strokes, bows or blows.
Of course, sometimes the attack of a sound varies too with the strength with which it is played and some samplers (Prophet 2000, Emulator II and Mirage) allow you to change the speed of both Filter and Amp Env Attack to imitate this effect. The Cello/Violin disk for the Mirage is an effective example of this technique.
Sequential's Prophet 2000 actually goes one step further on this score. You can actually move the point at which a sample playback begins according to the velocity of the key strike. So, if you have sampled the bow scrape of a violin or the tongueing of a flautist on a fast note, you can use a gentle key stroke to ignore this portion of the sample and start in the smoother sustain part of the sound (Figure 3). Coupled with the Velocity Attack Time, this can sound devastatingly realistic. It can also be devastatingly creative too. Why not apply this expressive element to a Hammond Organ sample (with or without key click) or a snare sampled with lots of reverb? The possibilities are tremendous.
Some sounds, however, vary so much between key strikes that even filtering and loudness can't cope. Here, most keyboards allow you to use two samples (one of a hard strike, the other a soft strike) and then either switch or cross-fade between them automatically according to the key velocity. The switching method gives you an either/or situation and can be a little drastic for some applications, whilst the cross-fade method is much smoother allowing a balance between the two sounds, although it can err on the side of subtlety. You should also beware, as velocity crossfade requires two samples to be played for each note and so this will halve the voice capability of your keyboard in this mode.
Again, you need not use velocity switching/cross-fading purely for imitation. Try creating unexpected combinations of instruments, particularly effective for solos or accompaniments. One of the best disks I have ever made up for the Emulator II featured not only Glockenspiel and Marimba but two new instruments - Marimspiel and Glockimba. You can hear Geoff Downes using this on 'Voice Of America' from the new Asia album.
The thing to bear in mind at this stage is that any facility the manufacturer provides to help you closer approximate an instrument can always be used creatively in lots of other ways, many of them more exciting, so don't be afraid to try things out. When you have used one Preset (or program) to build an authentic version of an instrument, use the next Preset to mess it about and turn it on its head!
The same is true of multi-sampling and overdubbing which we will deal with in the last part of this series, next time. This will describe how to make sounds work over a larger pitch range and how to build up an entire keyboard preset to turn the sampling device into a truly playable instrument. In the meantime, concentrate on getting your samples in tip-top shape, ready to start dotting them up and down the keyboard, as the best mapping out or keyboard assignment will not make one iota of difference if the sounds you use are duff to start with!
Feature by Paul Wiffen
Previous article in this issue:
Next article in this issue: