The final episode - Martin Russ closes this FM programming series with a thump!

So far in this series all the programming examples have been relatively straightforward - a few changes here and a minor modification there. The general purpose voice called 'Vanilla' has served us well and is a very useful starting point for any sound creation - the DX's 'Init Voice' has rather too many parameters turned off for my taste! In this final article I will attempt to tie everything together that we have explored so far. But first, a practical exercise in exactly the same vein - a few tweaks and we will have an orchestral thump!

When Ian Waugh reviewed the sounds I created for one of the DX7 ROMs in a past issue of Sound On Sound, he said that my 'LSO off CD' voice sounded 'distorted'. Exactly correct - I was in one of my rough and dirty programming moods at the time, and so the effect was deliberate. I have been called in by some people to roughen up their FM sounds - to remove some of the purity and delicacy of tone - especially by keyboard players in guitar-based 'hack and thrash' Heavy Metal bands. As in each of the techniques we have looked at, adding noise and roughness is just another aspect of FM. The original preset ROM sounds which accompanied the DX7 were a good advert for FM synthesis, but were only squeaky clean sounds. So let's get dirty!

ORCHESTRAL HIT

Stacks are the main weapon in our assault on this task. Rather than go overboard with a 4-Operator stack (as in Algorithm 1), I will choose Algorithm 4. This has two stacks of three Operators - quite enough to cope with for now, as well as a rather useful Feedback loop around Operators 4, 5 and 6.

The first thing to do is to analyse the required sound - readers who own a sampler should have a few classic cliched examples ready to hand, as they usually come as part of the standard library with most samplers. Drum machines, and even the Roland LA synths, contain examples of orchestral thumps as well, so your ears should have no shortage of source material. Ready with your analysis (Mr Spock)?

Depending upon which examples you listen to, there are really only three elements to the sound: a rapidly rising pitch sweep up to the final note, a short held note with full orchestra, and usually a percussive crash as well. A little reverb (or even a big one!) often helps to reinforce the effect, since it reminds your ears that the sound they are listening to is meant to be in a big room. Since my bedroom is not large and my reverb is an over-worked REX50, I usually revert to cheating and set long Release times instead (Rate 4s to the FM-literate out there!).

Pitch changes are new to this series. So far I have kept us firmly with both feet on the ground, so here goes with a final fling...

The Pitch Envelope can be used to do terrible things to the pitch of FM sounds, rendering innocuous voices unusable, but is also capable of subtle and very effective tweaks to the pitch. The rapidly rising pitch up to a held note which characterises the orchestral hit sound suggests that we should start the envelope below the Sustain level, but this makes it difficult to maintain the pitch when the sound dies away. Instead, a rapid rise in pitch, followed by a fall to the held note should suffice.

THE THUMP

Starting as usual with the 'Vanilla' voice and looking first at the Pitch Envelope, a simple change of Level 2 to 55 and a slowing down of the Rates to 98 should suffice. This produces a quick blip in pitch, which has a remarkable effect on the sound of the Vanilla voice. Since I am in a generous mood this month, I here include a free extra - the 'SOSpixie' voice. Just edit the Modulator Levels to 65, and you have a very unusual vocal sound, almost without effort. All I have done here is clear up the sound by reducing the number of harmonics, giving a purer sound.

'SOSpixie' isn't really an orchestral thump at all. So next a change of Algorithm to 4, as planned above, means that we need to re-shuffle two envelopes - Operator 4 is now a Carrier and Operator 3 is a Modulator. As an exercise, I will leave it to you to figure out how to swap them over.

The sustained envelopes of 'Vanilla' are obviously wrong for a percussive voice, so here are the Carrier and Modulator envelopes I threw together for the purpose:

I copied these across to Operators 1, 2, 4 and 6. Because I wasn't watching, the previous envelopes remained for Operators 3 and 5, and I didn't notice until later - this shows the perils of working too fast and not checking. You will find that I finally left Rate 1 for both of these Operators at 99, and altered most of the other Rates and Levels to much the same as the values shown above. The resulting voice has a slowish rise, which really brings out the pitch change, but is rather too mellow, especially since we are supposed to be dirtying the sound!

Turning to the left-hand stack (Ops 1, 2 and 3), I increased the Output Level to 90 and listened to just the relevant Carrier (Operator 1, if you aren't following!). I then shifted the Key Transpose down an octave to C2, to try and locate the sound in the correct register, and then set about creating a nasty sound. A Ratio of 1.00 to 1.00 all the way up a stack does nothing for unsubtle sounds, so I tried a few values for the Modulators, eventually deciding on 3.00 for Operator 2 and 4.00 for Operator 3.

There is no denying that this is an unpleasant sound. Turning off Operator 3 (you do keep turning Operators on and off, and then listen, don't you?) revealed the cause, so I quickly set the Output Level back down to 80, and added some Velocity Sensitivity in (2) to make the 'dirt' velocity dependent. The resulting sound was now too static, so I added judicious amounts of Detune to the Modulators to liven it up. At this point you might just be able to make out the sound we are working towards.

The other stack promises to be more interesting. First, set the Feedback to 3. Returning to the Brass sound described in Part 4 of 'Practically FM', you should remember that this helps to fill out the harmonics in a sound, and here that should be very useful. To roughen things up from the current state I increased the Output Levels to 90. The resulting sound has some of the tearing effect of the other stack, but needs more tweaking.

After a bit more experimentation I decided to retune to 2.00 all the way up, and add lots of Detune as well. The Carrier is detuned away from the other stack, so that we will create lots of 'beating'. The move up to 2.00 reverses the transpose down to C2, and gives a hint of shrieking violins!

Combining the two stacks restores the bottom-end to the sound - an example of using two stacks tuned an octave apart. Adding a bit of Velocity Sensitivity to Operator 6 reduces some of the brightness at low velocities, just as in the first stack. Our orchestral thump sound is now almost 90% complete - I just added a few changes of Rate and Level (it looks really suspicious if you have Rates and Levels all the same) in the envelopes and added a Rate Scaling value of 2 to Operators 2, 3 and 5, and a value of 1 elsewhere - just to thin out the high end of the keyboard.

4 OPERATORS

With four Operators the task is similar, but the simpler 4-Operator FM synths tend not to have any Pitch EG function, leaving us with just the raw sound - add a Pitch EG similar to the 6-Operator version if you can, since this improves things a lot.

As with the 6-Operator voice, a stack of three seems appropriate, and so I chose Algorithm 3, with the remaining Operator (4) tacked on at the side with the Feedback. My perverse nature means that rather than copy the 2.00/2.00/2.00 Ratio in the 6-Operator stack, in this case I decided that a 4.00/2.00/4.00 stack sounded better, especially because the other low Ratios from the other stack are missing.

The distinctive rough sound stems from the detuned Operator 4 with all the Feedback. A value of 9.89 sounds best to my ears, but do I need to remind you to experiment for yourself? The envelopes are similar within the constraints of their more limited possibilities, although I have forced the Carrier envelope to decay when the notes are held, unlike the 6-Operator version. OK - I will be honest. I did this because I didn't like the sustain sound, and this was the easiest way out. I am sure you can put it right by now if you need to.

If you have the extra waveforms in your 4-Operator FM synth (TX81Z etc), then you could improve the sound from Operator 4 by using one of the harmonically rich waveforms to better simulate a stack of Operators. I found that transferring the voice shown to a TX81Z from a DX100 produced a much brighter sound, so you may need to lower some Output Levels to taste.

That about finishes it. Not an exact sample re-creation, but a very useful sound nonetheless-try adding a bit of reverb and playing big block chords in a choppy fashion.

THE THEORY

Not too much theory this time - if you think about it, a full orchestra is going to have lots of harmonics, and the pitch change as they all come to the note. After that they decay to a more restrained sound. That's about all there is to it - a couple of stacks to produce lots of mush, and the pitch sweep to give it character.

FINAL SUMMARY

And so we come to the final summary. I have not had space to explore all the possible avenues of FM synthesis in this six-part series, and so I have chosen to give pointers to useful approaches and tools to using FM. Here are some of the main points I have stressed throughout this series (in no particular order):

- Detuning across stacks makes sounds natural sounding and less synthetic.

- Turning Operators on and off and listening helps you understand what's going on. It splits the sound into smaller bits.

- Experiment with values - try out the possibilities.

- Concentrate on Envelopes first.

- Remember the basics - Carriers affect volume, Modulators affect tone.

- Try out different Ratios and learn their characteristic sounds.

The overall message of this short series can be summed up in the following few words: Don't be overawed by the 'aura' of complexity which some people build around FM. Instead, take your time, keep your head and, most important of all, experiment with your ears!

I hope that you now have a better idea of how to approach FM synthesis - allow me to wish you 'good programming!' for the future. Let me know via SOS if you come up with any interesting sounds as a result of this series. Cheers!

I would like to thank David Bristow, John Chowning, Yamaha Kemble (UK), the X Series Owner's Club, Soundbits Software UK, Lawrence Wilkes of 5 Pin Din Software, and all the others who have helped me in my learning of the inner mysteries of FM.

I am working on a disk for the Atari ST which will contain all the example voices in this series, plus a few other interesting sounds from my own library, for both 4-Operator and 6-Operator DXs. This will be made available through SOS Shareware.

• The 6-Operator Dump shown is from Lawrence Wilkes' DXPERT program. A shareware version of this featuring just the Librarian functions is available for £7. Contact SOS Shareware, (Contact Details).

• The AI Editor Dump is from my DX7 Artificial Intelligence Editor V1.3. A Public Domain version with limited facilities is available for £7 from SOS Shareware.

• The 4-Operator Dump and Table is from the Soundbits DX100/27/21 Editor/Librarian program. There is also their 4X4 program, which can cope with all the Yamaha 4-Operator synths except the FB01. Contact Soundbits Software UK, (Contact Details).

Frequency Modulation synthesis, or FM for short, is based upon a different set of fundamental building blocks than conventional Analogue Synthesis. FM has one element, the OPERATOR - a device for producing pure sounds called sine waves, with an ENVELOPE GENERATOR used to control its output level. Operators can be connected together in structures called ALGORITHMS. Sounds flow from the top of an Algorithm to the bottom. The Operators at the bottom are called CARRIERS, and they 'carry' the sound into the outside world, whereas the Operators higher up in the Algorithm are called MODULATORS, and they 'modulate' or modify the sound.

In general, to alter the volume of a sound, you change the Output Level of the Carriers; to alter the brightness or tone, you alter the Output Level of the Modulators. To alter the overall shape of a sound, you therefore need to change the Envelope of the Carriers; and to alter the way that a sound changes tone with time, you need to alter the Envelope of the Modulators.

The pitch of a sound is usually associated with the frequency of the Carrier, whilst the basic tonal quality depends on the ratio between the frequencies of the Carrier and the Modulators.

It is worth remembering that FM synthesis uses only two fundamental ways of controlling the Operators: you can alter their Pitch or Frequency, and you can alter their Output Level. All the extra Scaling, Modulation and Velocity Sensitivity features do is affect either the Pitch or the Output Level of the Operators.

Finally, since we will be concentrating on 'live' editing of sounds in this series, rather than the laborious 'keying in' of voice patch sheets, you should ensure that you have read your synthesizer owner's manual thoroughly. Make sure that you are familiar with its front panel editing controls and their uses. This series is not designed for the beginner - the sections which explain FM in the owners' manuals are often very good for the person just starting on FM.