Fathom Synth Development Thread

[Kwurqx]

I mentioned Dexed and Oxe FM as open source FM synths earlier (I think). As examples of practical implementations.

Oxe FM is free, but not open source, AFAIK.

Oxe FM went Open Source in 2015 (as far as I know)
https://www.kvraudio.com/news/oxe-fm-sy ... urce-30983

[Vertion]

Click on Buy, click Buy On Seaweed Audio.
Second option is Fathom Pro Update (Password Only) for $2.50
Click on that.

https://youtu.be/sqgW-2orQQg

[jbraner]

braner, Go to the web site www.fathomsynth.com/home

Click on Buy, click Buy On Seaweed Audio.
Second option is Fathom Pro Update (Password Only) for $2.50
Click on that.

DOH! Sorry, I didn't see the update option. I only just bought this, and did grab the last update for free.
I haven't really used Fathom yet (it's a little deep for me) but it is worthwhile to support you

[fmr]

On the AM FM page of the first oscillator select the Ratio Multiply dial.
Set the Ratio Multiply dial to a whole number value such as 4.000.
Add a modulator Sequencer and make sure the mod type is set to multiply.
The Sequencer is easier to manage than an envelope for all flat segments.
For a Ratio Multiply max value of 4.0 we want the modulator to hit the following values.
For super clean FM we want: 1.0, 2.0, 3.0, 4.0.
For gritty FM we want: 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0.
We will choose the last case with a total of 7 levels.
Now here's the tricky part, we want the Sequencer levels to match up with these ratio values exactly.
Since the sequencer zero amplitude will be used we want the sequencer Quantize Amp + 1 = 7,
which is Quantize Amp = 6.
Now we create some steps in the sequencer.
It will sound great since we are jumping the ratio levels but they always land on precisions of 0.5.

Who said that FM sounds gritty when we have ratios other than integers? It doesn't have to. I think you didn't undertood FM yet.

One thing is the modulation ratio, another is the modulation AMOUNT. It seems that you are assuming that the modulation index is ALWAYS at 100%. IT ISN'T AND IT SHOULDN'T. That's where enters the envelope to control the output of the modulator. The envelope determines the modulation intensity, by controlling the volume output of the modulator, and therefore the intensity of the generated sidebands. Ratios other than the whole number or 0.5 produce partials that are not harmonic (inharmonic partials). These are very important to create a lot of different timbres. Synths are not just about saws and squares. Percussion sounds, for example, are mainly noise, adn have really noisy transients for the attack portion. The bow attack in the strings also produces mainly noise. The same can be said about the blow on the brass instruments or the string picking in a guitar.

These attack transients could never be accurately reproduced with subtractive synthesis, and that's one of the reasons the instrument emulations were never really convincing.

Another thing is that many instruments (most notably human voice), even in the body of the sound have a spectrum with many inharmonic partials. It's all these rich spectrums that can be recreated with FM when we use those "strange" modulation ratios, especially in combined modulation chains, with two or three modulators. Of course, the modulation intensity has to be carefully controlled, and that's why the envelopes and many other real-time modulating sources, like keyboard, velocity, modulation wheel, etc., are so important in an FM synth.

Also, in FM the chain is complex because most of the time the sound is divided into different components. So, for example, in an algorithm where we have three pairs of carrier-modulator, we will use one pair for the first attack transients (which usually very noisy), another pair for a second attack stage, softer, and the third pair for the sustained part of the sound.

Anyway, all this is really complex, and difficult to create in a modular environment. I would be glad if you just allow for feedback (modulation of an oscillator by itself) and independent control of modulation ratios in each modulator, also with the possibility of a fixed frequency. That alone would already be enough for the FM needs in a modular synth. As you said yourself, trying to recreate a DX7 in Fathom would be nonsense.

[Kwurqx]

Attached are 2 examples of FM Bass patches for Fathomsynth (programs, not audio).

Both are variants of the same simple 3 operator configuration: 2 modulators, 1 carrier. All 3 "operators" are Additive Spectrum oscillators.

One has some metally "electric bass" harmonics going on in the higher spectrum.The other is just a growly thing.

Expect nothing fancy. They are just meant as examples and a starting point for further FM (or AM) experiments in Fathomsynth for those who are interested.

Just fiddle some with the partials amplitude and phase. And Amount and Ratio's. Add carriers and/or modulators. Or parallel signal chains.

You can only use Detune, Noise etc on the Carriers, that are at the end of the FM (or AM) chain(s).

You can off course also use FM (or AM) to create a (more) static spectrum to send into filters. It's very usefull in that way too. FM (and AM) can create very rich spectrums.

GHK BS FM 2M 1C.Bass.Kwurqx._.0.Program.zip

[FathomSynth]

I'm reposting the video since it got lost on the last page and I want people to know how to modulate FM correctly.

https://youtu.be/IKnLHEW4Yfo

[mevla]

Hello,

I'm trying to buy the latest update but at completing step 3, 'Review and Purchase' it cycles back to payment and the paypal option. Already paid twice in a row today It does not get out of the step 3 loop. Is this a known problem ?

Cheers.

[FathomSynth]

That's the first I heard of that.

Which web site are you buying it on? KVR, Web Site or Shopify?

I will send you a PM with some instructions.

[FathomSynth]

This post is for FM experts only, it gets into Fathom's actual FM implementation, so please don't read it unless you have the FM background to understand the equations.

FMR, I went back and double checked my FM code.

I discovered that for Fathom’s FM I’m not using the standard FM equations by default.

However if you set Fathom’s “Amplitude Source” to Carrier, then it will use the standard FM equations.

I think the reason I did this was because I thought it sounded better to multiply by the modulator frequency twice. The good news is that setting Amplitude Source to Carrier will make it work like the FM on all other synths, which is a softer sound, and it will not alias as quickly as the Amount dial is moved up. This should allow users to stack oscillators more easily without aliasing. Sorry there isn’t more on this in the manual. I also wonder why I did not make this setting the default case.

Just to be sure I’ll post here the actual math I’m using in my code, and you can check it if you like.

This is the standard FM math used by Fathom when Amplitude Source set to Carrier which is the “softer” case used by most Synths. Though I initially wrote this from memory, I checked a couple FM academic papers to make sure it matches.

Carrier:

y(t) = sin(2*PI*fc*t)

Where fc is the carrier frequency after being modulated.

fc = fn * (1.0 + A(t))

Where fn is the frequency of the note, which is the original unmodulated carrier frequency, and A(t) is the amplitude of the frequency modulation. One is added so that when the index or amount of modulation is zero then the original note frequency will result.

The modulator frequency is a ratio R multiplied by the original carrier frequency.

fm = R * fn

Where R is the FM frequency Ratio. On Fathom R is two dials Multiply/Divide.

Therefore the modulator signal is given by:

x(t) = sin(2*PI*fm*t)

And the amplitude of frequency modulation is given by:

A(t) = I * x(t)

Where “I” is the FM "Index", labelled “Amount” on Fathom.

The simplest FM uses sinusoids for both oscillators, but the equations above are no less true for any waveform such as a wave draw having a function F(t)

y(t) = F(A*2*PI*fc*t)

Regardless of the FM output waveform complexity, the pattern will eventually repeat itself, and this is the resulting Period of the output. Obviously if the output period is related to the original carrier period by X or 1/X where X is an integer then the output will be very “tonal” and will sound like either an octave above or below.

If the periods are related by a ratio X/Y where X and Y are both low number integers then the output will be “harmonic” but not completely “tonal” on the same note. This result is based on the exact same principal as chords, and essentially you are using FM to create chord sounding output.

Obviously for irrational values of X/Y such as 37.3/17.9 the output would have an extremely long period, potentially in the seconds range and generally would be considered atonal by most humans.

Of course there could be certain species of Aliens that would find this tonal, as well as Whales.

Your examination of Fathom’s FM however caused me to stumble on an ingenious idea which I’m not sure has ever been done.

You could force the FM output waveform to remain tonal with the original note regardless of the complexity by simply chopping it off at one period (or two) and rounding off the end to zero like I already do with the frequency pulse oscillator and then sending the result through the FFT to completely stomp out any aliasing.

A phase control could also slide across the entire FM output period thus selecting which portion is selected for the one period window, and even this in itself would radically alter the sonic possibilities based on just one set of FM input parameters.

This would be relatively easy to implement, would require only a couple extra dials on the FM page, and would open up an infinite realm of FM possibilities for users, less familiar with FM than you and I, basically protecting them from any aliasing no matter what they did to R and I.

We could call it “Rational Frequency Modulation” mode, or “FMR”

[SparkySpark]

We could call it “Rational Frequency Modulation” mode, or “FMR”

You're amazing!

[Kwurqx]

@Fathomsynth

It was not my intention to divert all attention to the FM capabilities and quirks of Fathomsynth....
And away from really nice and usefull mostly "classic" subtractive synthesis oriented features.

But since we're on the subject...

The good news is that setting Amplitude Source to Carrier will make it work like the FM on all other synths, which is a softer sound, and it will not alias as quickly as the Amount dial is moved up. This should allow users to stack oscillators more easily without aliasing. Sorry there isn’t more on this in the manual. I also wonder why I did not make this setting the default case.

Setting the "Amplitude source" to "Carrier" and then using the "Amount" to modulate the carrier just sounded like "normal" FM. But somehow, so did "Modulator". The difference in the workings of the "Carrier"/"Modulator" setting I couldn't "fathom" (sorry, corny word joke). In my "FM Bass" posted earlier I actually use the""Modulator" setting.

But this "Amplitude" of the modulator thing is also why use of partials of the Spectrum Harmonic oscillators as modulators in FM and changing amplitude and/or phase of them yields usefull FM results. Effectively you're thus also able to change modulator amplitude(s) and shape(s) and therefore the amount(s) and characteristic(s) of modulation.

Changes in the number of partials, their rank, amplitude and phase for the carrier quickly results in more dramatic effects on the resulting sound/spectrum.

O, you can set/modulate the (overall) Phase on the Carrier seperately. Since carriers will retain access to other oscillators pages/parameters in FM/AM mode.

[Kwurqx]

[oops]
Edit became post..
[/oops]

[glokraw]

? Is there a Fathom midi learn, or plan for it? Didn't see a mention
in the manual...it would be fun learning the dry knob
of the new fractal reverb when sequenced sounds are playing,
there's a lot of range to sweep.

The new reverb is a nice replacement for the reverb
in the presets, like having a whole new plugin!
Happy thanksgiving, everybuddy!

[Sinthoid]

I had some ideas for some out there filter features, mostly in line with the idea of implementing analog qualities to the sound without modeling.

1. Filter unison, up to 8 instances of the filter in paralell with varied filter cutoff and reso, just the same as osc unison. So you could recreate the "Tom Sawer sweep" using any filter module out of the box.

2. Many filters in various synths (notably the korg ms20) have noise in the resonance circuit due purely to nonlinearity in the circuit. This sounds distinct from, say, noise modulation of a filter, very unique sound. I think such a feature would compliment fathom's per voice distortion nicely.

3. Here's a fairly normal one, filter drive. Imo what would work best is multiple stages of soft clipping both before and after the filter, with independant controls for each. Volume and frequency reactive behavior would be nice too (think like tube amp drive).

These were just some crazy ideas floating around in my head, I'm not dying for these to be implemented per se.

[FathomSynth]

Sinthoid, Very interesting ideas.

Kwurqx, Yes, I do plan on adding a lot more power for editing partial spectrums, and of course that would be a powerful tool as an input to FM.