FM7 tutorial

[SoundCrafter]

Pretty much what the question said. I've picked up a demo copy of Tracktion 2 (don't worry, I won't be leaving MP anytime soon) and it came (somehow) with a full VSTi version of FM7 (unfortunately, I can't find the DLL of the full VST to put it into MP.)  So I would like it if anybody had the full tutorial from the old forums. I believe there was a link? If there was that will also work.

Thx in advance
SC

EDIT: Changed title.

[LPChip]

http://www.computermusic.co.uk/tutorial/fm/fm1.asp

[SoundCrafter]

Thanks a bunch LP, I hope you don't mind, but I'm gonna repost this...CM's format is ugly.


When Yamaha's DX7 synth came out in the 80s, it caused a sensation, immediately taking its place in musical instrument history by being the first affordable synthesizer to use frequency modulation (FM) to create its sounds. It was also one of the earliest pieces of equipment to boast a truly sophisticated MIDI specification, and it could store a massive 32 user programs. It sold by the truckload, and although it may not be at the forefront of musical creation these days, at the time it caused a distinct shift in the centre of gravity of synthetic sound production.

One of the DX7's drawbacks was the fact that it was a bugger to program. Allegedly, nine out of ten DX7s coming into workshops for servicing still had their factory presets intact. FM programming was seen as an arcane art only understood by professionals rather than bedroom twiddlers.

Indeed, the new FM jargon seemed completely unfamiliar; gone were oscillators, filters, cutoff and resonance, and instead we had to get our heads round 'operators', 'modulators' and 'carriers' (or were they the same thing? we couldn't be sure), not to mention 'algorithms'. Also, programming a stonking sound through a tiny LCD with a bunch of pushbuttons and a single slider seemed like less fun than keyhole surgery.

The principles of FM synthesis are actually very simple, but are so different from sampling and 'traditional' analogue synthesis most that people found their minds reeling from any attempt to make sense of it. But it really isn't that complicated, honest...

Consider what a good old-fashioned analogue low frequency oscillator (LFO) does when applied to the amplitude of an oscillator: it just makes some parts of the signal louder and other parts softer. In other words, the electronic 'wiggle' coming out of the oscillator first  wiggles enthusiastically, then less enthusiastically. This is essentially the meaning of modulation in this context: to change a waveform periodically. This is easy to grasp when the LFO is oscillating slowly, say one cycle per second; the amplitude of the sound increases during the first half second, and then fades over the other half second. But what happens when the LFO is vibrating at around the same audio frequency as the main oscillator?

Consider a smooth sound such as a sine wave that, if produced theoretically perfectly, is a pure tone with no harmonics - a pure electronic beep. [See Fig. 1]

If a second audio frequency sine wave is used to modulate or control the amplitude of the first, as the modulating wave goes positive it will boost the amplitude of the sounding wave, steepening the curve; and when it goes negative it will tend to reduce the output of the first wave, thereby distorting the originally smooth sinusoidal curve and in effect introducing extra harmonics into the sound.

This is the exact opposite of the analogue synthesis method that starts with a buzzy sound such as a triangle or pulse wave and then subdues it with a filter that selectively removes harmonics, usually tending to smooth the waveform.

[Some la-dee-da about what to use. We're going to use NI's FM7 for this tutorial]



Fig 1. A sine wave-A "pure electronic beep"

[SoundCrafter]

Before we go any further, we're going to need this.

Example 1: Smooth Operator
After initialising the edit buffer, look at the Matrix view and see how operator F is connected to the output by a right-angled virtual lead that contains a box at the angle showing the level setting. F is connected to the output and therefore acting as a carrier, as you can see from the shot on the right.

Notice in the lower left corner that the default waveform is sine (this is also reflected in the Spectrum and Waveform displays at the upper right). If your audio connections and drivers are correctly adjusted, you should be able to hear the sine wave played with the default on/off envelope. Before we make things more complex, play the sine wave as high and low as you can and notice how it's clearly audible in the upper registers but sinks towards inaudibility the lower you go.

Position your mouse pointer a few pixels above operator F in the Matrix diagram and slowly drag up. You should see another level box spring into life with a new virtual lead connecting back into F. This represents feedback, or the operator modulating itself. As you increase the level you will see the spectrum and waveform gradually becoming more complex (and you'll hear it, too, when you play). When the self-modulation hits around 40, the waveform has transformed into an approximation of a sawtooth wave and the spectrum display shows a series of gradually diminishing harmonics. This is the essence of classic FM: starting with a simple wave and modulating it to increase its complexity: not a filter in sight. 





Figs. 2-4: Before and after feedback is applied: the original sine wave has transformed into an aproximation of a sawtooth wave, and the spectrum display shows a series of gradually decreacing harmonics.

[SoundCrafter]

Example 2: only connect

Now comes the real FM stuff. Initialise again and right-click on operator E to switch it on. You can switch between views of E's and F's parameters by clicking them. Connect E to F by finding the point that is vertically below E and horizontally to the left of F and dragging up (a bit elusive at first, but it quickly becomes second nature). F is still a carrier but E has now become a modulator, not heard directly in the sound except in as much as it affects F's output. You've just created a two-operator algorithm without knowing it! Experiment with various level settings for E; as you will see and hear, you can produce some pretty complex waves with just two operators. [See Fig. 5]

Next, leave E's level setting at about 40 and locate the RATIO parameter (left). Make sure you're looking at E's parameters, not F's. Ratio sets the frequency of E relative to F (assuming F remains on 1). In other words, if E's ratio is set to 2, it will oscillate twice as quickly as F, or an octave higher. If set to 3, it will oscillate at three times the frequency, an octave and a fifth higher, and so on. Click on the up and down buttons or drag-scroll the Ratio's number. Play around with this. You can make a fair approximation of a square wave with E's ratio set to 2 and output at around 30. What happens when you set E's ratio to 0.5?


Fig. 5: Creating a two-operator algorithm is quite painless

[SoundCrafter]

Example 3: the important issue of keyscaling 
Did you try the experiment in Example 1, playing as low as you can go with just a sine wave? It fades markedly in the lower registers, as the sine wave carries little or no energy in the form of harmonics which make low notes more readily audible to human ears (though whales can hear them, apparently, but then they don't go in for FM synthesis in a particularly big way).

A partially satisfactory solution is to push the carrier's level to maximum, then reduce the level of middle and high notes with keyscaling. After initialising, click on ENVELOPE to replace the Matrix view with the Envelope and Keyscale view. Keyscale essentially alters the operator's level according to how high or low you're playing. The yellow line across the top of the keyscale graph indicates the level for any particular note. Its default setting is maximum for every note, so drag the little square breakpoint opposite C3 down to about -20 and the one on the right to about -40 and you  should hear the sine wave is slightly more evenly balanced across the keyboard range. Even with F's output level on 100, this still produces a fairly feeble sound in the bass area. What's really needed is more modulation (ie, more harmonics) for the lower notes than for the mid and high ranges, so it's the modulator, rather than the carrier, that needs to be keyscaled. Switch on E as a modulator, as in Example 2, and experiment with its output and keyscaling. Try making something like what's shown in the illustration below. The lines connecting the break points can be curved by dragging on the dots. Extra break points may be added by right-clicking.



Fig 6: This keyscaling curve for modulator E will create a more even response across the keyboard

[SoundCrafter]

Before proceeding any further, we must clarify a few bits of basic FM jargon to give us the mental tools to understand more...

- An operator is essentially a digital oscillator with a built-in amplitude envelope.

- In FM, if an operator is connected to the output it's known as a carrier.

- If it's connected to another operator in order to change that operator's waveform, then it's called a modulator. Operators can be either carriers or modulators or, as we shall soon see, both at the same time.

- The pattern of connection between several operators is called an algorithm. (In FM7-speak it's referred to as a matrix.)

Example 4: envelopes

Just as in other methods of synthesis, the sound of an FM voice is determined by its envelope as much as by its waveform. Happily, FM7's envelopes are much easier to program than those of the original DX series; it's just a question of clicking and dragging. After re-initialising, select F and look at its envelope: straight up, straight across and straight down.

It's a gate, or on/off, envelope. It has breakpoints just like the keyscaler, so experiment with dragging these about. Moving the top left breakpoint to the right brings the sound in less abruptly by increasing the attack time. Dragging the lower right breakpoint to the right creates a more natural tail-off for the note by increasing its release time. Both attack and release slopes can be further shaped by dragging the dot in the middle of the line.

Now connect E to F, as in Example 2. As you increase the modulation, you'll notice a strange 'blip' as you release the note if the carrier (F) envelope has a longish release time while the modulator (E) still has a gate envelope and stops modulating the instant you release the key. Drag out the release portion of E's envelope and hear the difference.

For another variation, restore F's attack to instantaneous (vertical) and delay the onset of modulation by slowing down E's attack. Does this give you any ideas?


Fig 7: The abrupt cut-off of modulator E's envelope is audible in F's un-modulated release portion. For a more natural sound, drag out E's release until it more or less matches F's

[SoundCrafter]

Example 5: Stringbells 
Given the experience of constructing a few sounds of your own, you should now be better equipped to understand some of the more complex presets supplied with FM7.

Select patch 2, Stringbells, and have a look at the Matrix view. Operators B, D and F are connected to the output, so they're all modulators. Operator A modulates B, C modulates D, and E modulates F. Switch off the three carriers with a right-click: silence! Note that if a carrier is switched off, any operator modulating it has no effect, either switched on or off. You switch off the whole stack, so to speak.

Now switch on each carrier in turn, in effect soloing them, to hear what each one does. Clearly, B and F (hard-panned left and right) are the string parts of the sound, while D (centred) supplies the bell part. Further explore the patch by finding out what each individual carrier sounds like without its modulator. C provides D with an extra 'ting' in the lower reaches, A and E supply richness and their feedback settings are essential to the stringiness of the sound.

Study the relative ratios: the string parts are not only detuned relative to each other, they're also tuned down an octave (around the 0.5 mark), but the master tuning (on the MASTER page, oddly enough) puts the general  pitch back up an octave. The bell carrier (D) has a ratio of 2, up an octave, so it ends up two octaves higher. Which is kind of what you'd expect from a bell, innit?


Fig 8: A close look at the Stringbells preset reveals a few tricks

[SoundCrafter]

Example 6: Dyna Bass

Patch 7 is a fairly typical bright bass sound. As in the previous example, you can switch operators on and off to hear what they contribute. F is the sole carrier, modulated by B, D and E. B and D are in turn further modulated by A and C. D's ratio is 0.5, putting the whole sound an octave lower, while E provides more buzz, which gradually fades back into the basic D and F sound. A, B and C add the transient, the click at the beginning of each note.

Study each envelope carefully to see how this works; D and F are gates, the others all have decaying shapes. Little or no effects are used on either of these sounds. Find the Easy Edit page (hint: it's called EASY) and experiment with a few of the preset FX settings.


Fig 9: Dyna Bass is a fairly typical bright bass sound


Well, that's all of it, I think. Not quite as nice as the one on the old forums, but I try.

? Future Publishing 2006. All rights reserved

[residentgrey]

copy pasting goodness...mmm paste. Nice guide thanks a bunch!!

[SoundCrafter]

Hahahaha...mmm paste....Yeah. LP, any chance we can get this Stickied?

[rewbs]

De-sticky-ing. This thread is interesting but does not need to live at the top of the list.

[SoundCrafter]

No, YOU don't need to live at the top of the list! Actually, it fits nicely in this forum...Unfortunately, it seems Computermusic removed said tutorial...bummer.

[LPChip]

Which makes it even better that you've posted it that time True, we miss the images, but hey, atleast we still have the text