Help with FM

[trance_lucent]

I have two questions, the first theoretical and the second practical.

1. Trying to workaround Zebra limitation of FM-operator waveform to sine only (well, there are several other waveforms too, but they are not very useful). I remember that in the past I was able to create good sawtooth or square waves with sine oparators only. But can I use it for further modulation? Of course, not directly. But maybe I can modulate these operators(one modulating another) separately with another operator?

I mean:



Explanation: Let 1 = sine, 2 = sine. Let the result of 2->1 is sawtooth. And let the 4 has sawtooth wave like 2->1. So is 3->4 equal to 3->2->1, where 3 modulates also 1?

Of course, I have to test it in the csound, and I will, but can't right now.

2. Trying to achieve simple FM sawtooth with two Zebra operators (like self-modulation does), but can't, the result is more likely to filtered impulse train. I think this is because of the phase of the operators (they are cosine in nature, not sine). How can I workaround it? Using phaser inbetween?

[Sendy]

In my honest opinion Zebra isn't the best synth for this kind of thing. You can use a saw wave as the modulator easily enough but not the carrier, unless there's some super clever workaround anyone knows of. You also have to watch broadband signals doing FM because the FM part of Zebra aliases easily (though, in it's defense, the mirroring can be a pretty cool sound design tool).

In general I tend to find the sine and sine-like waveforms more useful for FM anyway, because broadband signals tend to go to fuzz and break up under FM - it's a synthesis which starts with simple signals and builds them up in complexity. So an already "full" signal doesn't really have anywhere to go.

Even FM8, which has classic subtractive waveforms like square and saw, only gives you heavily band-limited versions of these waveforms, because they're more musically useful (in general) and won't cause harsh sounds.

I do quite like the analog crossmod-style FM on subtractive synths, but Zebra won't go there. What you need is a good VA synth like Diva or Cypher.

[trance_lucent]

Thank you Sendy.

As for aliasing - currently I don't want to thought about it. And I also don't fear it very much, after all, many classic synthesizers had it (JP8000, classic Viruses, of course DX7).

And I don't need high modulation amount - especially with harmonically rich carrier, so aliasing won't be very high. As for other synth - I want to stay in Zebra because of its flexibility.

Currently I am experimenting with constructing complex modulator - using stacked detuned oscillators, modulated OscFX, filters and so on. So it isn't possible with Diva. The result is very interesting - gritty waveform with complex structure, many spikes, edges, reminding some of Virus spectral waveforms or some of Galbanum waveforms and many others, but it isn't static, it constantly morphing. And this is from using just sine carrier, and there are also some problems. If I set modulation to relatively small values (10-20), the waveform sounds great, but have prominent first harmonic. And if I set modulation higher, it starts to lose its gritty buzzy character and became more dirtier and bellish - like classic overmodulated FM. Not very pleasant. I think that sawtooth/pulse/triangle carrier waveforms would be better for low modulation values.

Also I think that analogue-style FM isn't very convinient for my task - these nice spikes and edges are result of negative phase counter increments (waveform is played backwards for a moments) which are occurs only when PM is used (may be I am wrong), and thankfully Zebra's FM is PM.

[trance_lucent]

Why I need this technique?
To say the truth, if I disable FM operators and play just my complex modulator, the sound (not waveform) is similar to that with PM. Well, it has slightly different character, but in overall it is very similar.
But it (without PM) has all disadvantages of usual stacked oscillators - especially the beating. Its amplitude can vary drastically due to phase cancellations.
And PMed waveform is much more stable in terms of amplitude, having much less beating, that is good for mixing, especially for bass sounds.
And one of variants how it looks:

[xalama qo]

I like maths like this...even I can understand it!

[xalama qo]

... the mirroring can be a pretty cool sound design tool).

Hi Sendy, what do you mean by mirroring? I haven't heard that term before.

[hollo]

... the mirroring can be a pretty cool sound design tool).

Hi Sendy, what do you mean by mirroring? I haven't heard that term before.

+1 ???

[trance_lucent]

I think mirroring = aliasing.

[xalama qo]

yup, Sendy is obviously referring to aliasing, but I'm just not quite getting the mirroring analogy. Unless she means like foldback distortion which AFAIK inverts the signal peak above the clipping threshhold.

But aliasing is less precise than that, it's a guess-timation of the frequency that should be generated, and therefore almost random, although the aliasing frequencies you hear on the high notes of synths are usually quite consistent per note, even though their relation to the audible aliasing frequencies of the neighbouring notes is totally harmonically out of whack!

But agreed wholeheartedly, aliasing is fun to play with as a creative texture!

[trance_lucent]

yup, Sendy is obviously referring to aliasing, but I'm just not quite getting the mirroring analogy. Unless she means like foldback distortion which AFAIK inverts the signal peak above the clipping threshhold.

Aliasing means that frequencies that are generated/sampled over the Nyquist are mirrored into the audio range.

Edit: and, yes, aliasing can be useful. For example, for generating inharmonic series for bells or high hats.

[Sendy]

Yes, mirroring in the frequency domain, as opposed to the time domain, which is foldback distortion, or "wave folding". Get a rolling spectrogram running, send a bright oscillator to an FMO and slowly crank the FM amount up, you'll see partials literally bouncing off of the roof or the spectrogram and back down. Sounds a bit like ring modulation.

You can really grunge out a saw wave like this, it sounds a lot more usable than other synths that I've heard aliasing on.

[xalama qo]

Ok, cool! Thanks for the education. What's a good vst for a rolling spectrogram, preferably free please?!

BTW, Sendy, i always enjoy your posts. You have some serious insight into the technical side of sound, and I've often picked up nuggets of gold in your posts which have helped me in my less scientific approach to sound.

So no sarcasm intended at all when i say thanks for the education!!

[pdxindy]

The XMF filter with its filterFM is a world of exploration unto itself!

[xh3rv]

Explanation: Let 1 = sine, 2 = sine. Let the result of 2->1 is sawtooth. And let the 4 has sawtooth wave like 2->1. So is 3->4 equal to 3->2->1, where 3 modulates also 1?

I think the frequency bands equate but not the amplitudes of those bands. Seems like the difference between DX7 and FM8 - the 3->2->1 && 3->1 isn't possible on DX7 algorithms but is on FM8's matrix. Kinda wouldn't take my word for it TBH (I mean I think I'm right but there's a lot better out there on the subject)

[Sendy]

Ok, cool! Thanks for the education. What's a good vst for a rolling spectrogram, preferably free please?!

BTW, Sendy, i always enjoy your posts. You have some serious insight into the technical side of sound, and I've often picked up nuggets of gold in your posts which have helped me in my less scientific approach to sound.

So no sarcasm intended at all when i say thanks for the education!!

No problem. I just like to talk about things I'm passionate about Really need to get a blog sometime.

For freeware sound visualisers there are s(M)exoscope which will give you a good waveform view, and Sonogram SG-1 for spectrograph action. For paid stuff I prefer IL Wave Candy but you have to buy it in a bundle of other stuff. It looks pretty, though.