Aliasing and hardware synths

[briandc]

Hi everyone.
as some of you may know, I've been working primarily with software synthesizers in recent years. And one of the big "issues" about them seems to be the issue of aliasing. And now I've begun to wonder, what about aliasing in hardware synths? Does it occur often, even in "high end" instruments?

I'd love to hear from anyone here who has something to say about aliasing with regard to hardware synths (analog, digital, whatever)
Thanks!

brian

[madtheory]

Definitely happens with the Roland V-Synth XT. As you might have guessed, it's got nothing to do with the platform (hardware, software, DSP, PC in a box, WHY) and everything to do with the maths

My Novation KSR is a lot less prone to it than the V-Synth XT, but it does happen at ridiculously high pitches with certain settings.

[meatballfulton]

Lots of early VAs exhibit aliasing. MicroKorg, AN1X, JP-8000, Nord Lead...

[HalloweenJack]

In theory every digital synth is prone to aliasing, and the fact that you hear it or not is due to a combination of factors, only analog, for what I know, doesn't have aliasing (but I could be wrong), so yes, even high end synths can alias badly (Roland JDXA has a bad aliasing with PWM)...this article explains aliasing in a clear way...

I have digital synths which have aliasing, the V Synth and the Evolver (the digital waves inside the Evolver have lot of aliasing), but actually I don't mind, apart from aliasing those synths have a lot more to offer and even when they're aliasing for me is part of their "character"

[gcoudert]

It was pretty bad on my JP50. The JP8080 isn’t as bad as the JP8000 but it’s definitely there.

[madtheory]

The aliasing in the Evolver is deliberate:
https://www.gearslutz.com/board/electro ... izers.html

[briandc]

Hi everyone,
thanks for your posts and links.
I'm surprised to see the number of digital hardward synths that have been known to produce aliasing. Would you say that such synths have had more or less success because of the aliasing? Was there ever much discussion in the past about the occurrance (in manuals or among users)? Today it seems everyone mentions aliasing when talking about audio quality, but I don't recall it ever really being mentioned back in the 80s..

brian

[meatballfulton]

Early samplers with low sample rates were the first instruments to exhbit aliasing. For instance, a sample rate of say 8kHz is low enough that sounds with any higher frequency content above 4kHz would cause aliasing. When the DX7 arrived, even with it's higher bit rate because of the way FM works, it was easy to generate harmonics above the Nyquist limit. The problem continued with VA synths especially when doing audio rate modulations.

The way around aliasing is usually to just increase the bit rate...at 96kHz, only frequencies above 48kHz would cause aliasing. I am not knowledgeable about various anti-aliasing techniques used in modern digital synths, but there are many softsynths available today that I have never heard aliasing artifacts from.

To briandc and HalloweenJack, aliasing is a digital phenomenon only. It is one of the things that analog fans deride in digital instruments.

[HalloweenJack]

To briandc and HalloweenJack, aliasing is a digital phenomenon only. It is one of the things that analog fans deride in digital instruments.

Thank you for pointed that out, I was not sure it was only in the digital world.

Well, I don't deride it, I've learned with my V Synth that there are occasions where you can use aliasing creatively, and actually I find the aliasing in the Evolver digital waveforms ("those" Prophet VS waveforms ) quite interesting, it's part of the character and the charm of the synth

[Ashe37]

I'm currently sitting down with a Novation Peak and need ideas how i might be able to make its digital oscillators alias. The New Oxford Oscillators on the Peak are supposed to not be able to alias within the audible spectrum so I want to try...

[meatballfulton]

Ashe, are you serious?

What sort of FM routings are possible? The easiest way to get aliasing I can think of is extreme audio rate modulations and then playing the highest notes so the sidebands go beyond the Nyquist limit. Based on what I've read about the Peak, I doubt that's possible...The Nyquist limit is 12MHz, way beyond human hearing

[madtheory]

Evolver aliasing is deliberate. I guess it's because it was a"flaw" on the Prophet VS. It wasn't really- they based that on the PPG wave, everyone likes the aliasing on that too. I'd imagine Wolfgang Palm was ok with that!

Don't conflate bit rate with sample rate. Two different (but related) quantities.

Notation Peak is not a VA, I doubt it will have aliasing?

@meatballfulton where is the 12 MHz from?

[Ashe37]

Ashe, are you serious?

What sort of FM routings are possible? The easiest way to get aliasing I can think of is extreme audio rate modulations and then playing the highest notes so the sidebands go beyond the Nyquist limit. Based on what I've read about the Peak, I doubt that's possible...The Nyquist limit is 12MHz, way beyond human hearing

Yes, i was serious. I've still got it right here, hence my asking. I was taking their 'aliasing free' as a challenge.

[Ashe37]

Evolver aliasing is deliberate. I guess it's because it was a"flaw" on the Prophet VS. It wasn't really- they based that on the PPG wave, everyone likes the aliasing on that too. I'd imagine Wolfgang Palm was ok with that!

Don't conflate bit rate with sample rate. Two different (but related) quantities.

Notation Peak is not a VA, I doubt it will have aliasing?

@meatballfulton where is the 12 MHz from?

The oscillators are not and do not claim to be analog. They are FPGA digital oscillators. Also, some of the oscillators are sampled.

[pflosi]

To briandc and HalloweenJack, aliasing is a digital phenomenon only. It is one of the things that analog fans deride in digital instruments.

You can "synthesize" aliasing in analog with audiorate PWM, like a class D amplifier Sample and hold can do similar effects, too.