Do different oscillators have their "own" sound ?

[piRoN]

I don't know, I've never done an A/B test. I usually find the fact that oscillators sound considerably different to be fairly convincing evidence that oscillators sound considerably different.

[Stab Frenzy]

No doubt about it. But some people don't believe it, so they have to have their AB test. Not that they could actually be bothered doing the AB test themselves, but what can you do?

[Stab Frenzy]

Okay then, here are some square waves:





https://soundcloud.com/worng/square-waves

Signal path is oscillator -> pittsburgh Outs -> RME Fireface UFX

Each individual osc was normalised to -6dBFS to make the people who care about such things happy. All tuned to the same frequency, if your eyes tell you differently then try to remember what defines a complete wave cycle. I only have four VCOs in my system these days, got rid of a bunch of them for digital ones.

I imagine people will look at these and think they look pretty similar, so they should sound similar, right? But when you look at a waveform you see a time/amplitude graph, but your ears don't measure time/amplitude, they measure the amplitude of all the frequencies instantaneously. If you want a graphical representation of what you hear you need a spectroscope, not a waveform. The important differences in how the waveforms sound are trapped in the little wiggles around the corner of the square and you don't have a very good view of them.

The PW controls (on the oscs that had one) were set to the centre position, as you can see that results in a different duty cycle for each osc. Differences! I might even do this for sawtooths too. Although the Mangrove doesn't have a sawtooth, I'll see what I can make that sounds kind of like one.

[Stab Frenzy]

And the sawtooths:





https://soundcloud.com/worng/sawtooths

Same methodology as with the squares, except in this case the Mangrove doesn't have a saw wave so I had to make a saw equivalent for the comparison. I did it by ear using the formant section, the result was a biased pulse kind of thing with a lot of high frequency content in the corner of the pulse.

Regarding the differences between the actual saws, it's all in the amount of ringing in the turning point of the wave.

So now that we've gone through the boring stuff like what do the vanilla waveforms sound like, how should we test for the differences between oscillators when they're doing their actual job of making music? How do we test sync and reset behaviour in an evenhanded way? How do we quantify FM behaviour? What about oscillators that have both linear and log FM like the uLFO? Where do I point to on a chart to prove my opinion that the VCOb and uLFO sound beefy whereas something like the Anti-Oscillator sounds gummy? Do through-zero oscs get their own section because they can travel forward through time?

[Broadwave]

Another example are the square waves on a Minimoog - they're derived from a few fixed 5% resistors to create a voltage divider to set what *should* be a 50% pulse to create your typical hollow sounding square wave... but what you actually get is a pulse that could deviate from 46% to 49% - hitting exactly 50% is almost impossible to get without modding and fitting a trimmer for perfect adjustment.

So, even on a single Minimoog, each the 3 "square" waves generally have slightly different harmonics - but the result adds to the character massively.

[madtheory]

Not that they could actually be bothered doing the AB test themselves, but what can you do?

Ah but that would preclude arguing about it on de tinterwebz

[GuyaGuy]

https://soundcloud.com/worng/square-waves

Yeah those sound different. But I kept waiting for the bass to drop.

[Stab Frenzy]

Hmmm, I just listened to those on Soundcloud and the compression messes up the sound a bit, you can still get an idea of the overall tone though.

Interesting that there hasn't been much discussion of this since some actual examples were posted, ie something to actually discuss.

[KBD_TRACKER]

Hmmm, I just listened to those on Soundcloud and the compression messes up the sound a bit, you can still get an idea of the overall tone though.

Interesting that there hasn't been much discussion of this since some actual examples were posted, ie something to actually discuss.

Thanks for the tests. The sonic differences for different oscillators with saw and square waves are as I expected tiny.
Outside of physics labs, and in ordinary studio or music making/playing conditions, I would expect them to be largely ignored.

My contention since beginning was NOT that all oscillators were exactly identical sonically (plse reread my earlier posts) because this would be a false statement.
Interestingly enough, sometimes in forums when one says 43, one gets soon attributed of and chastised for saying 430 ...

My contention was and still is that the sonic or timbral differences between oscillators were sufficiently small (particularly after post-osc. processing ie modulation, filtering, etc. and all the way to a synth output) to justify doubts about whether it mattered what oscillator are in a given synth.

In other words I continue to believe that if one switched oscillators in a synth with some other ones from a different manufacturer, the net sonic change at the synth OUTS would be barely distinguishable and musically meaningless or useless.

[Walter Ego]

In other words I continue to believe that if one switched oscillators in a synth with some other ones from a different manufacturer, the net sonic change at the synth OUTS would be barely distinguishable and musically meaningless or useless.

Well, it's funny...when you decide in advance what your outcome is going to be, it usually ends up being true.

[pflosi]

Clear case of selecting on the dependent variable

[haj]

Why don't you take some vintage synth with aux in (which is fed to filter) and some stand alone osc (like from some modular) and compare the final result (osc->vcf->vca) with original synth's osc and external one? Then it may prove the "negligible difference" will be actually so or not. I bet the latter.

[madtheory]

Ooh goalpost moving. Are you mixerman? ROFL!

[piRoN]

In other words I continue to believe that if one switched oscillators in a synth with some other ones from a different manufacturer, the net sonic change at the synth OUTS would be barely distinguishable and musically meaningless or useless.

You really haven't used many synths then. There are plenty of synths let down by mediocre oscillators (of the ones I've owned, the DSI Evolver and Yamaha CS15 spring to mind), and there are plenty of synths where the oscillators are a standout (my ASM-2 for example). I don't buy and sell gear on "kind of - sort of" differences.

Additionally, differences in timbre are usually exaggerated by additional processing such as VCA nonlinearity, filtering, and distortion, NOT minimised. In the same way that a bland resonant synthline suddenly gains enormous spectral dynamics when you overdrive it.

But whatever, you've made up your mind and no amount of evidence to the contrary is going to change it. As a great mind once said: "I'm afraid this conversation can serve no purpose anymore. Goodbye."

[Stab Frenzy]

Hmmm, I just listened to those on Soundcloud and the compression messes up the sound a bit, you can still get an idea of the overall tone though.

Interesting that there hasn't been much discussion of this since some actual examples were posted, ie something to actually discuss.

Thanks for the tests. The sonic differences for different oscillators with saw and square waves are as I expected tiny.
Outside of physics labs, and in ordinary studio or music making/playing conditions, I would expect them to be largely ignored.

But we haven't investigated sync behaviour, linear and exponential FM, through zero FM, even how a VCO behaves as it moves from one frequency to another. Not to mention how the different harmonic content of different oscillators gives a filter different material to work with.

But yes, for the purposes of making unfiltered drones, case closed!

I have a fair bit of mastering to do this weekend but if I get it finished in time I'm going to set up a nice big parallel modulation test with all four oscs being modulated at the same time by the same CVs and all recorded at the same time to show a few more differences. However as you can see from this pic of the oscs I'll be using they don't all have the same inputs and outputs. Which is weird for things that all sound the same wouldn't you think?