Question about FM synthesis in Thor vs other programs

[aRiver]

Always I've wondered about this in Thor: if you do FM through routing between a few oscillators and dial oscilator volume/fm to the max it will turn into noise. Similar situation is in Subtractor - FM dialer is often useless past 15%,if not less. If you take Ableton's operator or FL's Sytrus I find you have wider range of sounds at your disposal because sound doesn't just turn into shit as soon, even on maximum settings.

Hence I've never used FM in Reason.

To give an idea of what I'm talking about here is an audio example

VERY LOUD



First audio part is Reason, second is FL. Hear how last few seconds of the first audio are just noise.

There are slight tonal and volume differences, waveforms, except not a full square on the 3d Thor osc, are the same, but the main thing is that when second FM oscillator comes in thor sound turns into noise, sytrus doesnt. What is the best way to approach this?

Reason:

Reason example.png (625.73 KiB) Viewed 3183 times

FL:

FL example.png (156.6 KiB) Viewed 3183 times

[Benedict]

I'd be interested in knowing the exact technical reason but I have noticed that difference too.

The DX-7 was like Thor in that it was easy to get into Noise with too much modulation. Some modern FM synths that claim FM are very tame in comparison. My best assumption has been that in reality they only offer a fraction of the possible modulation depth to make the synth easier to keep in the 'sweet spot'?? Nice in some ways but feels like cheating in others.

Of course, they could be using a different set of maths as Yamaha's DX-7 & Casio's CZ-1000 were both actually forms of Phase Distortion but the results were different - not dissimilar to what you are describing???

[Oquasec]

I like how reason handles the multiple ways to do fm in thor.

[QVprod]

I'd be interested in knowing the exact technical reason but I have noticed that difference too.

The DX-7 was like Thor in that it was easy to get into Noise with too much modulation. Some modern FM synths that claim FM are very tame in comparison. My best assumption has been that in reality they only offer a fraction of the possible modulation depth to make the synth easier to keep in the 'sweet spot'?? Nice in some ways but feels like cheating in others.

Of course, they could be using a different set of maths as Yamaha's DX-7 & Casio's CZ-1000 were both actually forms of Phase Distortion but the results were different - not dissimilar to what you are describing???

Seeing as a few fm softsynths are able to load dx7 patches I'm not sure if the maths are necessarily different. The question would be if PX7 has the same modulation noise issue as Thor.

The difference between the dx7 and Casio being FM and PM but can two FM synths have different maths?

[Benedict]

Yamaha fudged a bit when they called the DX "FM" as it really is a form of PM - but not the same algo as Casio.

Clearly there is a difference between the way some "FM" synths handle things as while some have limited knob travel before the wave becomes really nasty, others need several Operators working together to achieve the same messy sound. I would assume that any that load DX-7 presets are the type tat get nasty pretty easily??

[aRiver]

Good to know I'm not the only one to come across that, because at some point I honestly thought I'm doing something wrong.

As far as I'm aware there are DX7 patches available for Sytrus and you get can get Yamaha sound out of it. I'll try to test out depth idea, maybe thor will sound about the same at lets say 30%.

[Oquasec]

To imitate a dx7 you'd need to look at this https://en.wikipedia.org/wiki/Yamaha_DX7

But if you wanna go farther than that you basically can do anything in reason due to how it's built [A layout that keeps producers producing]

[Marco Raaphorst]

Are you using sine waves for FM modulation? The most complex oscs are less controlable.

Must say I totally love Abelton Operator since it came to the market.

[alex]

Some modern FM synths that claim FM are very tame in comparison. My best assumption has been that in reality they only offer a fraction of the possible modulation depth to make the synth easier to keep in the 'sweet spot'?? Nice in some ways but feels like cheating in others.

I've made a quick test to check fm(*) modulation depth differences between Thor, Zero, Quad and FM4.

and in my book your assumption is spot on!
So, Thor turns out to be the "deepest" so far...

Each day I learn something new, thanks!

(*): Two sine operators, one modulating the other. And yes, it's Phase Modulation actually. RP Quad seems the only one that used the "true" name.

PS: also, the modulation intensity does not seems to be linear across the modulation range: on FM4 for example it becomes more "intense" towards the end of the knob range...

[Oquasec]

I like how propeller-head handles all their plugs tbh. I think I know where my money's going towards in audio production plugins after using it for so long. As if the stock devices weren't enough, they even have all of these digital versions of hardware which is laid out in such a perfect way to me minus the ability the add different colors for the piano roll's notes... Reason 10 is a mile away but I can see the progress.

Now to nut hug props some more, I will say again how much these plugins are unlike most of everything else on the market in how they work and how they are laid out :]

[aRiver]

It looks like sytrus reaches almost the same frequency range as rob papens - around 2k (all other things equal means around 60-70% in Thor matrix) and it will sound similar.

There are differences in processing of FM on different pitches. Especially it becomes quite obvious if pitch difference between mod and car becomes greater. Seems like Thor has grit to it on lower pitches, however when FMing on pitches with difference more 4 between carrier and modulator, sound becomes inaudible (in defense though, it is possible to do some musical sounds using FM osc even with high pitch difference). Sytrus does well on any pitch difference, and provides interesting metallic sounds even if modulator's pitch cranked up to x32.

Take away from this: they sound similar on the same setting but, at the end it's two different programs (duuuuh).

[selig]

I've had no problem recreating DX sounds in Thor, where possible due to limited algorithms available. There are even examples in the FSB, where I recreated a few of my fave DX sounds. So as far as I know they employ the same core FM technology.


Sent from my iPhone using Tapatalk

[Benedict]

Some modern FM synths that claim FM are very tame in comparison. My best assumption has been that in reality they only offer a fraction of the possible modulation depth to make the synth easier to keep in the 'sweet spot'?? Nice in some ways but feels like cheating in others.

I've made a quick test to check fm(*) modulation depth differences between Thor, Zero, Quad and FM4.

and in my book your assumption is spot on!
So, Thor turns out to be the "deepest" so far...

Each day I learn something new, thanks!

(*): Two sine operators, one modulating the other. And yes, it's Phase Modulation actually. RP Quad seems the only one that used the "true" name.

PS: also, the modulation intensity does not seems to be linear across the modulation range: on FM4 for example it becomes more "intense" towards the end of the knob range...

Thanks Alex

Great idea as I think you just proved a lot of what I have been thinking in this thread and my Reason Sound Challenge thread too - (Stock) Reason synths tend to have a far greater bandwidth which means they are brighter, at a cost of sounding more papery, whilst many of the big names make their synths sound warmer by limiting the upper harmonics.

Also interesting that in this thread many of the replies are OT. The same in the Reason Sound Challenge thread. It is almost like people deliberately don't want to viddy the overlarge mammalian in the enclosed vicinity - as if that is a danger for them in some way??

Eenteresting

[selig]

Some modern FM synths that claim FM are very tame in comparison. My best assumption has been that in reality they only offer a fraction of the possible modulation depth to make the synth easier to keep in the 'sweet spot'?? Nice in some ways but feels like cheating in others.

I've made a quick test to check fm(*) modulation depth differences between Thor, Zero, Quad and FM4.

and in my book your assumption is spot on!
So, Thor turns out to be the "deepest" so far...

Each day I learn something new, thanks!

(*): Two sine operators, one modulating the other. And yes, it's Phase Modulation actually. RP Quad seems the only one that used the "true" name.

PS: also, the modulation intensity does not seems to be linear across the modulation range: on FM4 for example it becomes more "intense" towards the end of the knob range...

Thanks Alex

Great idea as I think you just proved a lot of what I have been thinking in this thread and my Reason Sound Challenge thread too - (Stock) Reason synths tend to have a far greater bandwidth which means they are brighter, at a cost of sounding more papery, whilst many of the big names make their synths sound warmer by limiting the upper harmonics.

Also interesting that in this thread many of the replies are OT. The same in the Reason Sound Challenge thread. It is almost like people deliberately don't want to viddy the overlarge mammalian in the enclosed vicinity - as if that is a danger for them in some way??

Eenteresting

According to tests I've done, most (if not all?) of the Reason synths/oscillators are bandwidth limited, which is often the complaint about Reason, right?
Forgive me if I'm totally misrepresenting/misunderstanding your post, but just to be clear are you saying a) this is not true and b) full frequency response sounds worse (papery) to you than bandwidth limited response?



Sent from some crappy device using Tapatalk

[Oquasec]

Interesting. So basically stick to reason for sound design ok thanks for the info

[Benedict]

According to tests I've done, most (if not all?) of the Reason synths/oscillators are bandwidth limited, which is often the complaint about Reason, right?
Forgive me if I'm totally misrepresenting/misunderstanding your post, but just to be clear are you saying a) this is not true and b) full frequency response sounds worse (papery) to you than bandwidth limited response?

Maybe Bandlimiting isn't the right term but clearly, there is something that some synths (the "modern" sounding ones) do that seems to limit their freq range to effectively boost the body parts of the sound. Maybe you can offer us the correct term?

I was noticing exactly what Alex did that the Spectrum Analyzer showed a distinct difference between the outputs of e.g. Zero & Thor. Thor is strong right up to the limit of the window while Zero stops about dead somewhere around 2kHz. As I'm sure you would agree, less high freqs in a sound equals a greater perception of "warmth", more higher freqs can sound "papery" or thin in comparison due to masking. The Spectrum seems to support my theory that Reason stock units are very 'accurate' (for better or worse).

As a matter of interest, I made a simple piece (Kik, 303 & Synth String) in Bitwig using the stock Sampler, Polysynth & Kick synths and then the same in Reason using Thor, Sub & Thor synths and rendered off. Interestingly the spectrums (In Pot Player) showed the same lack of activity in the high freqs, despite the string in the Bitwig version being noticeably brighter than that in Reason.

Now I get the impression that some people are taking this subject as some sort of threat to something. It sure isn't. I am NOT trying to say that Reason, VST or any particular synth are better or worse. I have always wanted to understand why one synth seems to sound different from another. Sure the obvious answer is that the architecture (maths) is different but that doesn't necessarily answer how to get say Thor sounding like a Mini Moog off of a record from 1972 or a Wubbed up Massive from 2016 - both of which it should take a good stab at. Understanding what makes each system sound like it does (in a practical sense) is a goal of mine not only for my own music but because it keeps cropping up in fourms like this one.

From the latter, I assumed people would be interested in this but instead, it seems that the conversation is met with attempts to drag off-topic with tangential comments and hair-splitting that definitely seems intended to derail the true agenda. That is also an interesting (damned annoying & offensive) phenomena worth understanding as it seems that there is some unspoken idea that only certain things can be discussed in a society that otherwise champions itself on being inclusive & open minded.

[Oquasec]

After using reason's modules aloooooot more than usual I feel like...like I don't want to use vsts anymore.
This is how reason has me feeling right now :[

...And studio one is always there if I change my mind...might as well dig into this gigantic ass modular system some more to see what else I glossed over. To think the manual is just the beginning of how deep this daw is, which is insane given how despite being 1000+ pages long, not having this entire site's extra information in it. [Which is clearly past the 9999+ mark]

I have muuuuuch more to learn from this thing.

[kungubu]

I have also noted the difference in FM modulation depths on different synths - and it differs between analog oscillators too - but I don't think that can explain all the differences in sound. In the case of Thor you can notice how the lower frequencies gets attenuated as the modulation increases. Also you don't get the spacey metallic movement of frequencies when you increase the modulation slowly - it rather sounds like the sweep just adds harmonics. And when you modulate really deep (with a modulated modulator) you get more of a noise than the mayhem of chatter that often is the case when you modulate analog osc in the same way. When I make more percussive FM sounds in Thor I think it vworks fine, but when you want to do mor specey sounds and slow FM-sweeps or slow moving drones Thor hasn't workt at all for me. When I now compared it to the bento osc in Reaktor I can see that the modulation depth goes deper in Thor - the bento stops at 15k which is about 90% modulation in Thor. But if you do a FM-sweep with a modulator and a carrier (sines with the same freq) in linear FM all the way in bento and 90% in Thor the sweeps aren't remotely similar. One thing is that the bento don't attenuate the low frequencies. But another thing is that you in the spectrum analyzer can se how the harmonics move as they increas in the bento sweep as the modulation increases while you hardly se any movement, just an increasing amount of harmonics, when you do the same sweep in Thor (both in the analog and the wavetable osc).
I have just compared clean sweeps of FM amount in the bento osc and thors analog and wavetable osc and I must say that the bento osc comes far much closer to the sound of a similar raw FM-sweeps in the analog and digital modular oscillators I have worked with.

[alex]

Nice and interesting discussion is going on here... well, OT messages aside... (Benedict I hear you )

I just want to add some more observation, but first, please consider that I'm not expert so if I say something wrong, all kind of corrections and insults are welcome but please don't sue me!

As far as I can tell FM modulation "depth" depends on the modulator operator volume, so, more operator volume -> more FM depth.
Based on that, I believe that the only conclusion I could get from my quick video test (1Sine Oscillator -> FM'ing -> 1Sine Oscillator, both at C1 pitch/note)
is to confirm Benedict first post impression, that is, Thor when using FM via modulation matrix has more modulation depth (volume) compared to other synths(*), that should explain why it turns into "noise" early...

However, in my opinion this doesn't tell the whole story ...
The modulator oscillator sure can have more volume, but how much bandwidth it has? (this is - I believe - what Selig meant)
How much bandwidth the carrier gets (the modulated one)?
In other words, what is the frequency range they (and their harmonics/side bands/FM "overtones") can travel before aliasing or some other form of "bandwidth limiting" occurs?
How much bandwidth for the FM process itself? Does it happens at full audio rate (@project sample rate) or is it "capped" (downsampled maybe?) somewhere in the audio path, for cpu usage considerations?

Bear with me, I don't have the answers and I'm not even sure if all those questions make sense, I'm just thinking out loud here.
So, I very welcome anyone who has FM/DSP knowledge to chime in and shed some light on this "obscure" but fashinating topic

(*) to be completely fair, in Quad you can raise the modulator (Osc1) volume further than I did in my test, either via the dedicated Osc1 volume knob and/or via its mod matrix (constant -> OscX volume, which applies also to Zero) and thus increasing the modulation depth.

PS: interestingly enough in Thor the Osc FM type has less FM modulation intensity/depth compared to the mod matrix FM routing

[dioxide]

From what I recall the Thor FM Pair Osc is based on a DX7. The DX100 and DX21 both go further in terms of FM volume and the values on a DX7 and DX21 are not the same.

Thor's FM in the Mod Matrix is a different type of FM than that used in the FM Pair or a DX style synth.

[selig]

Here are some quick comparisons between Thor FM Pair and PM7.

Thor FM Amt = 1
Gives the same harmonics as
PM7 OP2 Level = 40

To continue
Thor @ 1 = PM7 @ 40
Thor @ 2 = PM7 @ 50
Thor @ 5 = PM7 @ 60
Thor @ 10 = PM7 @ 70
Thor @ 30 = PM7 @ 80
Thor @ 60 = PM7 @ 90
Thor @ 123 = PM7 @ 98 (best match)

So while they (surprisingly) end up being the same at both ends of the knob, it's that pesky 'curve' between those two points (where things often go astray!) causing the massive differences - especially in the low end of the controls.

[kungubu]

Here are some quick comparisons between Thor FM Pair and PM7.

Thor FM Amt = 1
Gives the same harmonics as
PM7 OP2 Level = 40

To continue
Thor @ 1 = PM7 @ 40
Thor @ 2 = PM7 @ 50
Thor @ 5 = PM7 @ 60
Thor @ 10 = PM7 @ 70
Thor @ 30 = PM7 @ 80
Thor @ 60 = PM7 @ 90
Thor @ 123 = PM7 @ 98 (best match)

So while they (surprisingly) end up being the same at both ends of the knob, it's that pesky 'curve' between those two points (where things often go astray!) causing the massive differences - especially in the low end of the controls.

Sorry for my ignorance, PM7 is?

[selig]

Here are some quick comparisons between Thor FM Pair and PM7.

Thor FM Amt = 1
Gives the same harmonics as
PM7 OP2 Level = 40

To continue
Thor @ 1 = PM7 @ 40
Thor @ 2 = PM7 @ 50
Thor @ 5 = PM7 @ 60
Thor @ 10 = PM7 @ 70
Thor @ 30 = PM7 @ 80
Thor @ 60 = PM7 @ 90
Thor @ 123 = PM7 @ 98 (best match)

So while they (surprisingly) end up being the same at both ends of the knob, it's that pesky 'curve' between those two points (where things often go astray!) causing the massive differences - especially in the low end of the controls.

Sorry for my ignorance, PM7 is?

Opps, big TYPO on my part. I meant to type PX7, not PM7!
;(

[kungubu]

Here are some quick comparisons between Thor FM Pair and PM7.

Thor FM Amt = 1
Gives the same harmonics as
PM7 OP2 Level = 40

To continue
Thor @ 1 = PM7 @ 40
Thor @ 2 = PM7 @ 50
Thor @ 5 = PM7 @ 60
Thor @ 10 = PM7 @ 70
Thor @ 30 = PM7 @ 80
Thor @ 60 = PM7 @ 90
Thor @ 123 = PM7 @ 98 (best match)

So while they (surprisingly) end up being the same at both ends of the knob, it's that pesky 'curve' between those two points (where things often go astray!) causing the massive differences - especially in the low end of the controls.

Sorry for my ignorance, PM7 is?

Opps, big TYPO on my part. I meant to type PX7, not PM7!
;(

So they changed the modulation curve between FMPair and PX7 - I always thought the oscillators were identical (but I've never used PX7). I guess the curve of the PX7 is closer to DX7. But he modulation depth seems to be about the same. Compared to the FM modulation through the mod matrix the modulation depth of the FM Pair seems to be quite limited. If you modulate a sine (130Hz) 1:1, @ 127 on the FM Pair seems to be about equivalent to @ 43 in a mod matrix FM modulation of two identical sines from two Analogue Osc. Full mod matrix FM will take the spectrum all the way beyond 20kHz. The strange thing is that the difference between the FM sound of the oscillators isn't that big – the analog oscillator FM sounds still very digital – and you don't really get any of the phasing sounds of pitch instability characteristic of analogue FM.

[avasopht]

I use the noisy parts past the threshold to give the attack portion a little kick (or for characteristic modulation).