Understanding the DX7 (Part 2)
The end of our quest to make Yamaha's revolutionary DX synths easier to comprehend and program. Jay Chapman performs the closing ceremony.
The final episode sees some new areas investigated and a few loose ends tied up. The rest is up to you.
It's my opinion that the velocity/touch-sensitivity feature on its own justifies the difference in price between the DX7 and the DX9, and is a major contributory factor in putting the DX7 into a class of its own when compared to the other polyphonic synths available in the same price range.
The use of touch-sensitivity as applied to carrier Operators should be obvious, since the effect is exactly that obtained when playing with more or less strength on a piano keyboard. Those synth players who haven't had much experience of piano playing will find a whole new world of musical expression opening up before them - though at the expense of having to learn some new technique. Personally, I'm happier with the action of the DX7 keyboard than with that of a real piano, but if you aren't, there's always the DX1.
The DX7 goes further than this simple amplitude effect, however, since it's also possible to alter the harmonic content of a sound depending on how hard a key is struck. The ability to vary harmonic content via touch-sensitivity is an important factor in the synthesis of acoustic instrument sounds. Consider the different tones you get from striking a note on the piano (or picking a guitar string), gently and then heavily - the difference, which may well be quite subtle, is not merely one of volume. Consider the 'slap and pull' method of bass guitar playing, which is a more exaggerated example of the same string played in different ways to produce sounds with very different harmonic content. To be fair, the above is an over-simplification - the 'pull' on a bass guitar string gives harmonic movement (both actual and subjective) for various different reasons, including the way the human ear reacts to the fast initial attack and decay. Using touch-sensitivity on both modulating and carrier Operators allows realistic synthesis of such sounds.
A good first voice to investigate for the effect of touch-sensitivity on carrier Operators is ROM 1A Voice 9 'PIANO 2', which uses algorithm 18 (shown in Figure 1a). Since there is only one carrier Operator, we can try the different values (0 to 7) of the Key Velocity Sensitivity (KVS) parameter without any complications caused by other carriers. Go into Edit mode, press the relevant green keypad (it has '28' on it) and select Operator 1 with the purple Operator Select keypad. You should keep the DX7 volume control high during this exercise so that you can hear the KVS effect clearly.
With the parameter value at 0 there is no touch-sensitivity, while with a value of 7, you have to strike the keys inordinately hard to get any sound at all. Most of the time, a value of between 2 and 4 should be used for the carriers, otherwise the effect becomes a little unmanageable. Where a voice uses more than one carrier, the KVS values can be set differently for each, so that striking the keys with different strengths alters the mix of the sounds. In other words, we get different relative amplitude levels of carriers for different key velocities. We'll see this in action in the next example.
The other possibility is to apply touch-sensitivity to the effect of the modulating Operators by effectively varying their output levels (as we've just discussed for the carrier Operators) which alters the harmonic content output by the modulated Operator. This latter Operator need not be a carrier but may also be part of a vertical stack of Operators - experiment with all the possibilities.
Typically, KVS is applied to both modulators and carriers to allow the musician to add plenty of 'feel' to the performance, since both the mix and the harmonic content of the components of the mix can be varied via keyboard touch. If you examine ROM 1A Voice 11 'E. PIANO 1' which is a pretty convincing imitation of a Fender Rhodes piano, you'll find both carrier and modulator Operators being affected through KVS. 'E. PIANO 1' uses algorithm 5 (shown in Figure 1b), which provides three sound components via the Operator pairs 1/2, 3/4 and 5/6 whose amplitude mix is controlled via carrier Operators 1, 3 and 5. Further investigation shows that Operator 1 is responsible for the amplitude of the 'Fender Rhodes Tines' part of the sound, and this Operator has a KVS parameter value of 2, compared to Operator 5 which has a KVS value of zero. So, the harder you strike the keys, the more apparent the 'tines' component will be in the overall mix (relative to the Operator 5/6 component at least - you can clarify this effect by toggling Operator 3 off).
Finally, if you look at Operator 2 you'll see that it has a KVS value of 7, which has a (very) subtle effect on the harmonic content of the Operator 1/2 component. Try varying the KVS from 7 to 0 with Operators 3 and 5 toggled off: you might think such a small variation is of little consequence, but you'd be wrong - such subtlety is the secret not only of accurate imitative synthesis but also of giving your own voices character.
This section groups together all the obvious controllers, as well as some that are perhaps not so obvious! They are: the pitch bend wheel, the modulation wheel, after touch, breath, modulation footpedal, volume footpedal, poly/mono, portamento and the data entry slider.
Most of what follows is simply advice based on my own experience - less in the way of explanation is needed since the controllers do not present the same theoretical problems as the principles of FM synthesis.
Four of the controllers are governed by the four identical sets of parameters titled Range, Pitch, Amplitude and EG Bias (green keypads 17 to 32 in Function mode). Pitch and Amplitude simply govern whether or not any Low Frequency Oscillator modulation is to be applied to Operator pitch and amplitude - see E&MM August '84 for full details of LFO routing.
The Range and EG Bias parameters warrant a little further comment before moving on, however. The Range parameter, as its name indicates, sets the range over which the relevant controller has effect. At one extreme we have a very wide range (RANGE 99) which you'll find rather difficult to control with any accuracy, whilst at the other end (RANGE 20, say) the maximum effect available is subtler but far more controllable. I often find that the wider range is best kept for special effects while some smaller range is more suitable for general performance.
After Touch can be used to control the timing and amount of vibrato. This is useful when the left hand is playing chords and so cannot get to the modulation wheel: I find it 'feels' better too! By way of experiment, try setting up a suitable voice so that it will respond to pitch modulation (see the LFO Routing mentioned above if you don't know how) and set up After Touch to control the vibrato. Press the brown FUNCTION keypad followed by the green '30' and '+1' keypads to get After Pitch set to On. Now press the green '29' keypad so that you can play with the After Range parameter value via the data entry slider. If you push the slider right up so that RANGE=99 and then try to play, you'll find that the slightest extra pressure and the vibrato sounds like somebody strangling the voice! All right, I am exaggerating slightly, but if you try the Range value around 30 I think you'll agree it becomes a lot easier to use. Also, there's less chance of vibrato coming on by accident when you're playing the keyboard with different strengths using touch-sensitive effects.
I mention the pitch controller at this point because exactly the technique we've just discussed for After Touch applies to any controller that has a Range parameter. In the case of the pitch controller, this is particularly significant because the human ear is very sensitive to pitch changes, so the need for accuracy is great.
Many of the pitch bends used for musical purposes (as opposed to specifically dramatic effects) on a synth can mimic the bends a guitarist uses. Due to the width of the guitar fretboard and the fact that strings won't stretch infinitely, these bends are usually restricted to about one-and-a-half tones, and are more often than not simply one whole tone. On the DX7, we can use the pitch bend wheel, set at the full RANGE=12, to bend up or down a whole octave, which I'd put into the dramatic class of bends! Trying to bend just one tone accurately, especially in the heat of a performance, is an almost impossible task with Range set to 12, so why not step it down to 2, which gives a whole tone bend?
The data entry slider, and the '-1' and '+1' keypads can be used during performance if it makes sense. For example, if you need to bend octaves during the chorus of a song and whole tones during the verse, don't leave the Pitch Wheel Range set to 12 and thereby mess up all the whole-tone bends. Instead, select the voice you're going to use in the normal way and then press the brown Function keypad followed by the green '3' keypad so that the data entry controls can vary the Pitch Bend Range parameter.
When you want octave bends, simply push the slider right up, while for whole-tone bends push the slider right down and then hit the green '+1' keypad twice - that way you don't need to fumble about with the slider trying to see whether or not a '2' has come up on the display!
Another possibility with the data entry slider is perhaps best explained by actually doing it. Select ROM 1A Voice 6 'Strings 3', get into Edit mode, and select the Frequency Coarse parameter on Operator 1. Play chords with your right hand and move the data slider up and down: it's a sort of cross between a stepped filter and somebody singing 'wah' into a vocoder. Good isn't it? No? Oh well, tastes do differ!
You can also switch into Edit mode (having selected a voice) and then use your knowledge of how the sound works to let you 'toggle off' some of the Operators to switch part of the sound in and out. Of course, you could simply keep the two different versions of the same sound in adjacent voice slots, but the 32 internal voice memories can soon all get used up in a three hour set. Buy a RAM pack - or alternatively, a microcomputer!
I'll let you experiment with leaving other Function or Edit mode parameters 'open' to be changed by the data sliders - you'll find that some are useful, others less so.
The Breath Controller, as used for 'blowing' sax, harmonica and brass voices like those found among the Yamaha pre-programmed sounds - their names end with the letters BC - needs a lot of practice before you'll get really good results. It can be well worth editing the supplied voices to make more use of this controller. As an example, try altering the ROM 2A Voice 5 'SAX BC' to give more of a rasping sound when you blow over hard, thereby imitating the same effect on the real instrument. To get you started, try increasing the Output Level of Operator 3 to around 85 and its Amplitude Mod Sensitivity (green keypad '16') from 2 to 3. You'll find that the increase in output level more or less balances the attenuation introduced by the sensitivity change until you blow fairly hard, at which stage Operator 3's output asserts itself. I picked Operator 3 because it has a feedback loop and was therefore likely to produce the 'harsh' component of the voice. This change is too crude as it is - I would continue editing the voice (perhaps by playing with Operator 2 in a similar fashion) so that the rasping sound is fuller and less buzzy.
Some nice effects are possible by using the Breath Controller (or After Touch if you don't have the BC) to give independent control of part of a voice. The obvious example is ROM 2B Voice 19 'E.P-BRS BC' which sounds like an electric piano until you blow, at which point a brass sound comes in. This is useful for sharp rhythmic brass 'punches' or slow brass crescendos over fading piano chords, for example. With careful use of Keyboard Level Scaling, it should be possible to separate piano and brass and play clean brass fills (ie. without the piano sound 'doubling') over piano chords.
All I want to say about the Poly/Mono and Portamento sections is that you should experiment carefully to make sure you understand the possibilities available. In particular, the use of Portamento with chords controlled by a footswitch can give dramatic results, and different Portamento effects under each of 'Poly' and 'Mono' should all be investigated. The Voice Library with Performance Notes sheet that comes with the synth is very useful and gives plenty of guidance in this area, eg. ROM 1B Voice 31 should be played in Mono mode with some fingered portamento to give a fretless bass sound.
Having read through this series of articles you should have seen by now that the DX7 is not the mysterious and uncontrollable beast that some reviewers (having made no attempt to understand its use) make it out to be. I hope that I've succeeded in liberating a few DX7 (and DX9) owners from the misconception that it will forever remain a 128 voice preset synth in their hands.
This series is necessarily incomplete in the sense that there are many possibilities and combinations of factors that have not been touched upon. The articles were intended merely to set the ball rolling, so to speak, and with luck, the hints and pointers given will have set you off on your own exploration of what is a marvellous musical instrument.
This is the only part of this series active so far.
Gear in this article:
Feature by Jay Chapman