Practically FM (Part 3)
Part 3 of our series for FM synth owners, in which Martin Russ attempts to give away even more secrets of FM programming. Operators at the ready!
Martin Russ reveals more programming techniques for FM synths this month as he attempts to recreate a Dave Greenslade electric piano sound.
What was the important thing about Electric Pianos? Why were they so popular in the '70s? What would you concentrate on if you were programming an electric piano sound? If you thought the answer was the famous 'tine' sound, that tinkly, metallic plink that is a cliche all of its own, then in my opinion you would probably be wrong!
What actually made the Fender Rhodes and Wurlitzer pianos so popular amongst musicians was not the poor imitation of a real piano, or the unresponsive velocity sensitivity. No, what made them special was their ability to sound different. You may think that all electric pianos sound much the same, but do they? Think about the distinctive sounds of Supertramp, Greenslade, Soft Machine, Herbie Hancock etc. In fact, the remarkable thing is that if you know the music then a couple of chords is enough for you to determine who it is. If this sounds like the sort of advice that you give to aspiring guitarists - 'develop your own sound', then it has many parallels.
Times change and the electric piano seems to have forever gone from its main role in life - it always seemed destined to be supporting a monosynth or string machine as I remember. Even so, although the instrument may pass out of popular usage, there is always a place for a successful formula. The current success of Stock, Aitken and Waterman may cause some of us to remember the halcyon days of Chinn and Chapman, and in a similar way the synthesizer substituting as electric piano is certainly well capable of use and re-use in a variety of today's musical forms.
There are already a large number of electric piano voices for the DX7. It is probably fair to say that the E.PIANO 1 voice on the original ROM nailed the coffin for the electric piano. Anyway, not flinching from the task in hand, I hereby present yet another electric piano sound-alike...
Greenslade, formed by Dave Greenslade (who does the music for the BBC's 'A Very Peculiar Practice') and Dave Powell in the days when synthesizers were new, exciting and horrendously expensive, were one of the first exponents of the 'mostly keyboards' progressive rock. The electric piano featured very heavily in their work, albeit often deliberately distorted and finger-twistingly fast, and it is this distinctive sound which I will try and move towards this month. It is an interesting observation that in FM we have a synthesis tool capable of such fine grained detail, but perhaps I should save self-congratulation until after I have shown you how!
As usual I will start with our 'Vanilla' voice (see accompanying patch). Nearly everyone uses Algorithms 5 or 6 as the foundation for electric piano sounds, so let's try Algorithm 12 to see what it throws at us. If we were trying to create a Fender Rhodes type piano, I would now use a high ratio like 11.00 or 13.00 in a Modulator for a 'tine' sound and develop the sound from there. But a Wurlitzer type sound suits our purpose much more, as well as being more unusual. The characteristic we want adds a sort of whacky, very non-piano 'wang' to the sound when we play hard. Difficult to describe but it sounds wonderful. I have already mentioned that we want some sort of distorted or overdriven feel to the sound.
Algorithm 12 gives us a pair and a 'three onto one' stack of Operators. I will use the Operator pair first to create a 'bare bones' electric piano sound and work from there into the finer detail. First, replace the envelope on Op 1 with a straightforward percussive envelope, very similar to the one we used last month for the piano voice but with rather more fast decay early on (R2):
R1 R2 R3 R4 L1 L2 L3 L4
85 85 35 50 99 90 00 00
Copy this envelope to all the other envelopes. The resulting sound would win very few prizes so far. Now turn off all the Operators except Ops 1 and 2 and let's concentrate on the basic sound.
I set the Feedback to 7 to roughen up the sound but it still wasn't metallic enough for my taste. So I tried altering the tone by changing the Ratio of the Modulator, Op 2 - a value of 2.00 was about right: nice and neutral, whilst 3.00 sounded rather too synthetic to my ears. I tweaked the Output Level of Op 2 up a bit to 80, just to brighten the sound slightly, and a Detune of +3 for Op 2 was somewhere between not being audible and being rather too much, which +7 definitely was.
To finish off this part of the sound I added a slight amount of Velocity Sensitivity (2 to Op 1 and 2 to Op 2). Remember that in this electric piano voice we want the velocity to change the tone and not just the level, so here both the volume and the tone are changing. Rate Scaling makes an important contribution to realism, and a value of 4 sounded OK. You can apply the same scaling to all the other Operators in this case.
Turning now to the second part of the sound, we have to consider the difference between a single Carrier/Modulator pair and this arrangement of one Carrier shared by three Modulators. As you might expect, the effects of the three Modulators add together when changing the Carrier, and it is wise to listen to each of the pairs alone (and the four Operators as a whole) as we add parts. Op 3 requires some Velocity Sensitivity to provide volume changes with slightly more emphasis than for the other Carrier, and a value of 3 felt about right to my fingers.
I mentioned that we wanted a distinctive sound effect when the notes were played hard, and so I will now build it from the three 'pairs' of Operators 3 and 4, 3 and 5, and 3 and 6. The sound I want has a sort of low frequency 'growl', so try setting the Ratio of Op 4 to 0.50 and increase the Output Level to 99. This sound should decay slower than the main sound so that you feel it come in after the notes are played, and so changing R2 of Op 4 to 60 and R3 to 50 should help make it linger. The percussive envelope I defined above decays to 90 first, then down to zero, but I felt that forcing a decay to zero immediately would create the sound I wanted and so I set Level 2 of Op 4 to 0. I said that this part of the sound should only be present at high velocities, so a Velocity Sensitivity of 7 seems to keep it under control. Switching Ops 1 and 2 on as well and playing the sound reveals that all seems fine so far, except that the 'wang' sound is still only half-formed.
I will use the pair formed by Ops 3 and 5 to deal with this. A Ratio of 4.00 sounded metallic enough to my ears with an Output Level of 85, especially when I restricted it to the higher velocities by setting the Velocity Sensitivity to 6. The resulting sound from all the Operators except Op 6 sounds a bit lacking in sparkle and brightness but remember that we have yet to add the final Modulator for Op 3, and the three Modulators add together in their effects.
In order to brighten up the sound I left the Ratio of Op 6 at 1.00 but increased the Output Level to 95. This gave a very synthetic sound which should blend well with the other parts when I start the final tweaking. In the previous two examples in this series I arrived at more or less a finished sound without any recourse to fine-tuning, in this case I thought I would try and improve on the result so far.
The final polishing of sounds usually consists of just a few modifications to the fine details of the sounds. Detune is probably a good place to start here. Unlike the String example, where the detune was an important element of the sound, here any detune will only be used to add a bit of movement to the sound. This is one case where detuning the two Carriers will work, and so I detuned Op 1 down to -1 and Op 3 up to +1. Operators 4 and 5 sounded better when detuned to -7 and +2 respectively.
Turning Op 6 on and off whilst playing showed me that it was rather too strong at the top of the keyboard, so I used the major tool for fine-tuning to correct it: Level Scaling. I used a relatively harsh amount of scaling: 50. Applying this with the negative Linear curve starting in the middle of the lowest octave seemed to be about right (I set the Break Point to G1). Something about Op 5 kept nagging at me and I eventually changed L2 to 50, although the difference to the sound is only marginal - but this is fine-tuning after all! And finally a bit of non fine-tuning! If you look at the accompanying Electric Piano patch sheet, you will notice that the Rate Scaling for Op 6 is 3, whereas all the other Operators are set to 4. This is merely an oversight in my programming, and since it doesn't noticeably alter the sound I have left it in. This is known in the trade as an 'artifact', since it plays no part in the sound and could be a remnant of a previous voice.
The resulting sound is certainly different to most of the other electric piano sounds I have heard on the DX7, most of which concentrate on squeaky clean sound-alikes of the instrument rather than a particular final sound. I love the effect of playing harder using this voice, especially when you add a chorus effect to the sound. The final voice, naturally, is called 'SOSelpiano'.
The 4-Operator FM synth owners may already have noticed that there are no Algorithms on their instruments which let you modulate a Carrier with more than two Modulators. I could use the usual Algorithm 5, but I would prefer to use a two Modulator version: Algorithm 4 seemed to be a reasonable choice. The resulting voice shares some of the same sound as the 6 Op version when played hard, but I couldn't resist using Op 4 (high up at the top of a three Operator stack with feedback) and so the final 4-Operator electric piano sound has the familiar glassy element as well.
FM voices usually consist of a couple of Operators doing most of the work towards the final sound, and everything else is used for adding the fine detail. This is why 4 Op sounds are noted for their lack of detail in comparison with the 6 Op versions. Just listening to Ops 1 and 2 in our example shows very effectively where most of the electric piano sound is derived from. Of the 'three onto one' stack Ops 3, 5 and 6 make sense since this is where the 'playing harder' sound I wanted is produced.
The sound produced by the combination of Ops 3 and 4 seems misplaced - it is a quickly decaying FM chiff sound. In fact, this is quite an important part of the overall piano sound - try removing it and then replacing it. As you will notice, it adds the roughness to the sound when you play harder.
Removing it makes the sound much more synthetic - it is in fact a deliberate introduction of a rough, partially noise-like element into the sound, and it helps enormously in making the final result sound realistic.
We have now covered some of the most important techniques you will need to know in order to programme effectively in FM. As in many fields, knowing how to approach the problem is often all you need in order to make progress, but not knowing what to do at all can stop any advance completely. These are the techniques we have been using:
All the time we have been working on a sound I have constantly referred to the contributions made by parts of it. Turning Operators on and off so that you can hear their effect is one of the most useful tools you have at your disposal in FM. It is much easier to work on several simple sounds than on one very complex one. This is the reason why Yamaha put the Algorithm diagrams on the DX front panel, because you need to be aware of Carriers and Modulators and the shape of the sound you are working on.
In every case when setting a Detune or a Level, I tried the available range of values and zeroed in on the value that sounded most right. By knowing what surrounds the sound you can more easily make alterations as the sound develops. Also, using relative judgements is much easier when you can compare it with other events.
The principle that changing the parameters of a Carrier affects the overall volume or pitch, and that changing the parameters of a Modulator affects the tone colour, is worth remembering and applying. This enables you to home in on the particular velocity sensitivity or scaling that you need to alter to have the desired effect. FM need not be complicated, but the control we have over it can be! The basics usually still apply even to the most complex sound. Lower the Output Level of a Modulator and it will always make the sound less bright, unless you have a Fixed Frequency lurking in there somewhere!
Finally, don't be afraid to investigate and try out ideas - it's free! There are a great number of example DX voices out there for you to fiddle with. Probing around voices by turning off Operators and listening to how the sounds are constructed can be very useful. See how the sounds are made up - in most cases you need only look at the Algorithm, the frequencies used for the Operators and their Output Levels, and the general form. Explore and enjoy!
The VoiceMaster DX21/27/100 editor for the Atari ST is available from Soundbits Software UK, (Contact Details).
A shareware version of Lawrence Wilkes' DXpert editor for the Atari is available for £7 from SOS Shareware, (Contact Details).
Feature by Martin Russ
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!