FM algorithm cheat sheet

How to make that sound...
KVRist
136 posts since 18 Mar, 2012

Post Wed Jul 28, 2021 10:52 pm

Is there a table that, for each algorithm, shows a list of sounds for which it's suitable?

User avatar
KVRist
443 posts since 4 Dec, 2019

Post Sat Jul 31, 2021 9:27 am

:?: I wouldn't know what you're trying to get exactly.
It does not become clear to me from your descriptions.
Nothing can be better than liberty and free time to spend as you choose. Don't let those modern slaveholders ruin your life!

KVRist

Topic Starter

136 posts since 18 Mar, 2012

Post Sat Jul 31, 2021 12:32 pm

juno987654321 wrote:
Sat Jul 31, 2021 9:27 am
:?: I wouldn't know what you're trying to get exactly.
It does not become clear to me from your descriptions.
Let's say i want to emulate a trumpet. I look up trumpet in the table. No entry, but i find brasses. In the 'brasses' entry, they suggest the algorithm number 199, for example. So i know that if i want to emulate a trumpet i could use the algo 199.


User avatar
KVRist
443 posts since 4 Dec, 2019

Post Sun Aug 01, 2021 2:29 am

In the first video at around 6 min. e. g. they say something that kind of goes into the direction of what you may have in mind. But I think it's not that simple.
I think a string instrument e. g. could actually be made in many different ways using different algorithms (or arrangements of operators which is the same here). Or you could use the same algorithm with many different settings that lead to similar results (?).
So they made certain instruments for the DX7 using certain algorithms but that doesn't mean that this was the best or the only way to do it back then.
Anyways, there may be such a list or a good overview for the DX7 instruments that maps to the algorithms THEY used. I don't know if it exists, though. It might be helpful to have it for learning purposes, though.
But then it would be much more interesting and informative to know what the individual functions of the individual operators was and why they chose that kind of algorithms for certain instruments over other algorithms, then.

Now, there are complete overviews of the algorithms from the DX7 at least as you may know.
Still, those algorithms could always be enhanced (e. g. just by adding more OPs) and they were just a selection they chose from all the logically possible ways in which you could arrange operators given the limitations they had back then.

So maybe by "algorithms" you just refer to the very specific historical ones they chose for the DX7 because generally speaking I think there's no limited set of algorithms for FM really. You could just arrange operators freely nowadays as I understand it? But I don't understand much about FM yet.
Nothing can be better than liberty and free time to spend as you choose. Don't let those modern slaveholders ruin your life!

User avatar
KVRist
443 posts since 4 Dec, 2019

Post Sun Aug 01, 2021 2:37 am

Another way of doing it would be to check presets in a DX7 emulation which are named "trumpet" and you'll find the concrete algorithm popping up if the emulation is a good one?
Then you could just check this algorithm, draw it on paper,
try to find out yourself how they made the sound,
and finally check again how they actually made it
if that's what you're looking for?
Nothing can be better than liberty and free time to spend as you choose. Don't let those modern slaveholders ruin your life!

User avatar
KVRAF
13027 posts since 8 Mar, 2005 from Utrecht, Holland

Post Sun Aug 01, 2021 5:56 am

FM is deceivingly simple. One sine modulates the frequency of another which produces harmonics that weren't there. With more complex algorithms (especially once feedback gets used) the results get chaotic and essentially unpredictable.

With analog synths you could take the approach to start selection with an oscillator, add some filtering and envelopes and call it a day. Not so with FM. I'd go so far and call FM patch designers wizards - it's almost a black art.

There are some often used approaches, like use some operators for the sustain part of a sound and the others for the attack/transient part of the sound. You need suitable algorithms for that, but once you start cooking and looking at what recipes others use, you'll spot those.
We are the KVR collective. Resistance is futile. You will be assimilated. Image
My MusicCalc is back online!!

User avatar
KVRist
443 posts since 4 Dec, 2019

Post Sun Aug 01, 2021 10:38 am

Yeah, it's much about learning by doing as with other methods of synthesizing sounds but this one could even take much more time to actually get somewhere. FM seems really hard to predict and that's the reason why I've been pushing off learning how the patches in Dexed etc. really work - again and again. I'll have to do it one day! But I always feel that there are more important aspects to sound design that one should learn first like sampling techniques I'm currently exploring. It's incredible what kind of sounds you could get if you sample a sound and use a good sampler and its filters/effects (like good reverb) or layering possibilities to get a completely different sound in the end. I would have never imagined...

I'm sure exploring FM will be fun one day but it's just very very difficult to find a real good starting point since there are only few good (!) tutorials on FM out there. It's black art for people who are both really advanced in sound design and who like to spend hours and hours experimenting and trying to understand how otherwise arbitrary mechanisms could still be controlled and somehow "cracked" and thus lead to sounds that won't actually just hurt your ears. It is possible to use FM techniques very very carefully and purposefully I believe but this will take you a lot of time to figure out just exactly how. Studying and trying to understand the more nice sounding patches from the DX7 would be my personally preferred way of doing it but there's just not many teachers out there who will explain you how these patches really work and what each operator really does. That's why I find it difficult to kind of "get my foot in" there. Sorry, but really diving into FM is something I would personally keep for last and bother about other things first. I appreciate your idea of seaching for such overviews as you have in mind there, though. That might also help to get started.
Maybe some specialist who does FM a lot will post something useful of that kind if we're lucky and if such helpful material exists at all? I'm not too optimistic that it does, though...
Nothing can be better than liberty and free time to spend as you choose. Don't let those modern slaveholders ruin your life!

User avatar
KVRist
443 posts since 4 Dec, 2019

Post Sun Aug 01, 2021 10:46 am

But there is something fascinating about those wizards who know how to use this kind of black art I must admit. Just like we're being fascinated by magicians and their slide-of-hands tricks. :D
Nothing can be better than liberty and free time to spend as you choose. Don't let those modern slaveholders ruin your life!

User avatar
KVRAF
4172 posts since 7 Jun, 2012 from Warsaw

Post Sun Aug 01, 2021 10:57 am

specially once feedback gets used) the results get chaotic
This, unfortunatelly.

But! F'em has this great feature of oscillator sync, which allows to regenerate the pitch even from messy waveforms.
http://djwarmonger.wordpress.com/
Tricky-Loops wrote: (...)someone like Armin van Buuren who claims to make a track in half an hour and all his songs sound somewhat boring(...)

KVRist
142 posts since 18 Jan, 2014

Post Sun Aug 01, 2021 11:22 am

try this Fm tutorial,for synthesizing real instruments (pdf download) at
http://javelinart.com/fm-synth-programming.html

KVRist

Topic Starter

136 posts since 18 Mar, 2012

Post Sun Aug 01, 2021 12:39 pm

The first video shows that there isn't a one to one mapping between an algorithm and a particular sound.

The second video makes me think, like also BertKoor points out, that complex algorithms rather than only determining the color of the sound itself, are used to add features. So perhaps just few operators are enough to characterize a basic timbre.

Ratios are also important: https://www.reddit.com/r/edmproduction/ ... _harmonic/


FM_Synthesis_of_Real_Instruments.pdf seems to be good, though i needed to quickly create a few generators for testing some software i'm developing.

User avatar
KVRAF
2006 posts since 8 Jun, 2018

Post Mon Aug 02, 2021 12:44 am

bitwise wrote:
Sun Aug 01, 2021 12:39 pm
The first video shows that there isn't a one to one mapping between an algorithm and a particular sound.

The second video makes me think, like also BertKoor points out, that complex algorithms rather than only determining the color of the sound itself, are used to add features. So perhaps just few operators are enough to characterize a basic timbre.

Ratios are also important: https://www.reddit.com/r/edmproduction/ ... _harmonic/


FM_Synthesis_of_Real_Instruments.pdf seems to be good, though i needed to quickly create a few generators for testing some software i'm developing.
mapping between an algorithm and a particular sound...

algorithms, yes the call it that way, in way they are, but in the case of DX7, you had 32 'algorithms', or better routings, for carries/modulators. i think it are 32..

Dexed or FM8 or F'em can do, i am not certain about Dexed, any routing you want.

it was a way to make it easier, i guess... you must look at the routings, which sine oscillator is a carrier or a modulator, and a carrier can modulate itself, of course.

if there is any correlation between routing and a particular sound...

mainly ratio's and envelopes determine it, of course with use of a well choosen routing.
so it is the routing of what modulates what.... and of course a partical routing works better for a certain sound...

FM is easy, but can become very complex, and in FM soft synths, you have a lot of overview and control about routing, ratio's, also independent frequency, the level (also important), the envelopes! never forget how important envelopes are in FM, in any synth, to create a particular sound.
take me as apple as reed-son as flesh-icon, immediately
shits the wise man. hero and blind. take me as
stick as tongue as dog... Music & Soundscraping

yul
KVRian
1190 posts since 26 Sep, 2002 from Montreal, CANADA

Post Thu Aug 19, 2021 8:31 am

Its simple. Just use an oscilloscope and try some basic stuff :

A)
One operator with feedback (saw wave)

B)
Then chain two operators 2>1>output
and try some basic ratios :

1:1, 2:1, 3:1 you will notice basic waveforms appearing.

Integer ratios are easy because they will give you fixed waveforms.

Everything in between gets more movements and dissonance.

Return to “Sound Design”