Anti Aliasing...how does it sound??

HanZen · Post by **HanZen** » Thu Aug 21, 2003 3:05 am

Yo,

well...my question is a quite simple one. I read on forums that some synthesizer suffer from anti aliasing....especially software synthesizers (VST). But somehow I have no idea how it sounds. I mean I know what anti aliasing is technically (I have to as an electrical engineer

)...but I don't know the audible difference between an aliasing free oscillator and one which has aliasing.

Maybe some sound samples anyone?:)

Hanzen

kensuguro · Post by **kensuguro** » Thu Aug 21, 2003 3:21 am

good question. I was wondering about the same thing too. Ok, so aliasing is when you get the stairstep effect, even though you want a smooth line. You antialias it to make is as smooth as possible. So I'd guess anti-aliasing is good, as in a graphics sense.

From what I know, I am guessing that the more aliasing, the more "bumpy" signals, the more unwanted harmonics. In contrast, the more anti-aliasing, the more smoother, the more "pure" the signals.

But judging from the way "anti-aliasing" is used in HanZen's sentence, you "suffer from anti-aliasing". And I'm wondering, how can you suffer from anti-aliasing when it's supposed to be good? So do I have it the other way around? Can someone clarify?

[ This Message was edited by: kensuguro on 2003-08-21 04:24 ]

visilia · Post by **visilia** » Thu Aug 21, 2003 3:28 am

If well implemented, anti aliasing shouldn't have a sound at all

Now seriously, the sound of aliasing could be described as something between a ringmodulator and mp3 encoding artifacts.

If you want to hear it yourself, just load a basic waveform, like a sine or saw wave (you could use the excellent MS-20 waveforms, provided by Bosone in the Modular forum) into whatever softsampler. Spread the sample over the entire keyboard and play some high notes. Now you should hear some disharmonic, ringmodulator-like tones. That's aliasing.

edit: Ah, now I see your post too, Kensuguro. If he really meant anti aliasing, it would be harder to answer indeed. A badly designed anti aliase filter could make the oscillator sound dull and it changes the phase relation between the harmonics, which results in a slight change of the timbre of the oscillator.

Cheers,
Vincent

[ This Message was edited by: visilia on 2003-08-21 04:37 ]

kensuguro · Post by **kensuguro** » Thu Aug 21, 2003 3:33 am

that looks like an example of nyquest mirroring tho.. The mirroring of the high harmonics at the nyquest value (half of sampling rate) is what causes the disharmony.

Isn't aliasing closer to bit quantizing, but in a much more minute sense?

visilia · Post by **visilia** » Thu Aug 21, 2003 3:42 am

Ehmm... As far as I know aliasing is the mirroring effect around the nyquist frequency. It's samplerate related. I thought it had nothing to do with bit quantization noise.

_________________

[ This Message was edited by: visilia on 2003-08-21 04:42 ]

kensuguro · Post by **kensuguro** » Thu Aug 21, 2003 4:00 am

aha! thnx for clarifying! Now I get it.

HanZen · Post by **HanZen** » Thu Aug 21, 2003 5:21 am

ah yes...sorry kensuguro,

of course oscillators suffer from aliasing (and not ANTI aliasing). Anti-Aliasing is a feature to prevent wave bands from coming closer to each other (Nyquist, Shannon). And I think that's what programmers want to achieve. Hard to explain in english

But ok...so oscillators which are not aliasing free cause disharmonics. Ok, I 'll try that later.

Thx