Do hardware VA's alias as badly as software?

[Urs]

BTW.: what are porkies?

porky pies = lies (rhyming slang)

Ah... yes. It's pretty safe to say that your statement is correct.

[Music Engineer]

All digital synths I tested in the past year or two have relatively aliasing-free oscillators and quite audibly aliasing filters

actually, there's a simple solution: just use a linear filter!

[Urs]

All digital synths I tested in the past year or two have relatively aliasing-free oscillators and quite audibly aliasing filters

actually, there's a simple solution: just use a linear filter!

Yes... a linear filter with a fixed frequency. No modulation whatsoever! That'll do the trick

[hakey]

It's pretty safe to say that your statement is correct.

No doubt the argument is "inaudible aliasing = alias free". Being charitable, one might say that that's stretching the truth rather than an outright lie.

[Music Engineer]

Frankly, I can't imagine how fast are modern FPUs - might be very realtimish for say 10-20 harmonics

sure, 10-20 are no problem whatsoever. but for a 50 Hz saw with spectrum up to 20 kHz, you would need 400.

[Shabdahbriah]

BTW.: what are porkies?

In Seattle's east-side, it is another term for "date".

[hakey]

for a 50 Hz saw with spectrum up to 20 kHz, you would need 400.

My ancient cpu manages to run Reaktor Prism, which allows up to 200 harmonics per voice, without too much trouble.

Perhaps fully additive is the way to go. Wouldn't it allow for alias free Osc FM (seems to be one area in which digital synths fall down)?

[Urs]

It's pretty safe to say that your statement is correct.

No doubt the argument is "inaudible aliasing = alias free". Being charitable, one might say that that's stretching the truth rather than an outright lie.

Well, ok, I'll admit I'm overly picky.

Just don't feel that claims like "aliasing free oscillators" are justified if the filters alias like hell afterwards.

[hakey]

It's pretty safe to say that your statement is correct.

No doubt the argument is "inaudible aliasing = alias free". Being charitable, one might say that that's stretching the truth rather than an outright lie.

Well, ok, I'll admit I'm overly picky.

And I'm really not that inclined to be charitable - simply anticipating the arguments that some might use.

Just don't feel that claims like "aliasing free oscillators" are justified if the filters alias like hell afterwards.

yep

[Music Engineer]

My ancient cpu manages to run Reaktor Prism, which allows up to 200 harmonics per voice, without too much trouble.

not bad but still not enough for the lowest registers. with 200 harmonics, you can have a full (up to 20 kHz) sawtooth spectrum for fundamentals down to 100 Hz. on the other hand, the point of modal synthesis probably is not in creating an alias-free sawtooth spectrum (you would need an alias-free impulse-train as exciter anyway - so the whole thing is kinda recursive). you may want to generate sparser spectra, in which case 200 may well be sufficient.

Perhaps fully additive is the way to go. Wouldn't it allow for alias free Osc FM

generally, additive allows free composition of any (possibly time-varying) spectrum. the problem in creating FM spectra would be shifted into the computation of partial frequencies and amplitudes for the additive synthesis engine. and this computation would become unmanagably messy pretty quickly. even a simple one-carrier, one-modulator (both with sine waveforms) FM configuration requires evaluation of a whole bunch of bessel functions for the partial amplitudes. now imagine a whole bunch of "operators" connected with feedback-FM and possibly non-sinusoidal waveforms. i'd probably not attempt to calculate partial frequencies and amplitudes in this setting.

[Aroused by JarJar]

So the look-up table method, does this, as AJJ suggest, simply involve sampling a waveform at a single frequency and then repitching it, with the undesired effects (missing harmonics for low pitches aliasing for high) following on from that?

Are there any modern VA's that use such a crude method?

I was not suggesting such a crude implementation, although it has been and probably still is used. I specifically referred to "reading wavetables". Large wavetables read with slick interpolation can sound beautiful. A single wavetable will still have to be a compromise between richness in the lows and aliasing in the highs. But it is not necessary to use a single wavetable- many wavetables can be mapped over note numbers or pitch ranges.

I suspect that many hardware synths use nothing more fancy than this. It is easy to test if this is likely the case. Play the highest note that does note that does not alias. Now set your pitchbend to +/-24 and go down two octaves from the highest note which does not alias and pitchbend all the way up. Do you hear aliasing? If you do, then there is a good chance that this (mapped wavetables) is system being used. There are other ways this might happen, but I think this is the most likely explanation.

[mkdr]

Whats your problem kid?

Do we now have to experience everything personally in order for us to make acceptable statements.

I heard if I jump from the 5th storey window its likely to kill me, Im going to have to test that theory as it cant be true.

That is my problem. It is LIKELY going to kill you, is not the same as it is DEFINITELY going to kill you. And the aliasing topic is so filled with misinformation and lies that there's no point in discussing it without any kind of actual data. If for once at KVR we could stick to the facts. That's all.

But still no one has any idea what the aliasing on synths like Nord Lead, Virus, V-Synth is.. And if it even is actually aliasing or something else.

Now this is insulting. I am studying DSP and I know what aliasing is, and I've had enough of experience with the mentioned boards (I don't need to OWN THEM, I tried them out in the shop with headphones and pushed them far enough to alias) to confirm they indeed are aliasing at one point or another.

I had no intention of insulting you. But I'm no mind reader either. I can't know what you know to determine what you actually know. So i can only trust on samples you can deliver.

And seriously. I'm interested too.

[EvilDragon]

I have already provided one audio snippet displaying aliasing on Kurzweil PC3.

[V0RT3X]

Sure must suck for the people who can't stand aliasing. Their loss I guess

[hakey]

generally, additive allows free composition of any (possibly time-varying) spectrum. the problem in creating FM spectra would be shifted into the computation of partial frequencies and amplitudes for the additive synthesis engine. and this computation would become unmanagably messy pretty quickly.

Okay. Isn't audio rate modulation unmanageably messy whatever method is used? Seems it's the last bit of analogue behaviour that digital synthesis can't quite replicate properly.