Oversampling?

[ferez21]

I have always overlooked this term, although in theory I know how it works - have tried tweak this setting on some effect plugins, but couldn't hear any noticeable differences (Maybe because the global project sampling rate is 44,000 ?).
Am I missing something? Can this setting make a noticeable difference with instruments? Haven't tried it yet on any instrument because it chokes my CPU.

[izonin]

It depends on your age. If you're, say, over 25, you won't hear much of a difference.
Oversampling effectively extends the frequency range and isn't required if you work at 96KHz.

[penguinfromdeep]

It depends on your age. If you're, say, over 25, you won't hear much of a difference.

Aliasing doesn't manifest only at nyquist. It can be audible in the baseband too. You may notice it or not, really depends on the material. The aliasing becomes usually apparent with more drastic settings like distortion etc and when you play notes in the higher frequency range for example.

[ferez21]

It depends on your age. If you're, say, over 25, you won't hear much of a difference.
Oversampling effectively extends the frequency range and isn't required if you work at 96KHz.

But I don't work on 96khz.....should I ?

[izonin]

It depends on your age. If you're, say, over 25, you won't hear much of a difference.
Oversampling effectively extends the frequency range and isn't required if you work at 96KHz.

But I don't work on 96khz.....should I ?

As I said, it depends. Young listeners' ears are more sensitive to high frequency content and will spot aliasing where we won't.
So there's your answer: if you produce music for a teen audience, it would be safer to turn the oversampling on. (x2 would be enough)

[e-modic]

So there's your answer: if you produce music for a teen audience, it would be safer to turn the oversampling on. (x2 would be enough)

Seriously, where is the sense in doing that? We produce music at highest standart, max sampling rate, expensive converters etc. and in the end 99% of the listeners will have your music as a 192 bitrate mp3 on their mobile devices. Congratulation audio revolution

[Gamma-UT]

It depends on the synth and what you're doing with it. This is a pretty extreme example. I've picked on Sonic Academy's ANA because it can generate some serious aliasing, and how it's not something that only people will bat-like hearing will notice because, as penguinfromdeep points out, it reflects way down into the audio range.

Synths that offer oversampling often do much, much better on the aliasing test, but there are situations with Strobe, for example, where I'd use oversampling. Others not so much. It really depends on the sound – fast modulations on higher pitches tend to bring it out. Oversampling will seem to make the high end less fizzy and a bit more defined. Sometimes the extra fizziness is what you want.

[penguinfromdeep]

Excellent example Gamma-UT! Very easy to hear the aliasing even on laptop speakers

[vespesian]

Oversampling is also useful with compressors/limiters (i.e, catching stray transients).

[bbaggins]

"...have tried tweak this setting on some effect plugins, but couldn't hear any noticeable differences"

You probably didn't hear any difference because there was none. That, or the difference was so minute as to be inaudible.

Aliasing isn't necessarily audible if it's below -60 dB, and most of the time it's well below -80 dB. It's not cumulative, so even if you have low-level aliasing on many tracks they usually don't add up to audible levels.

Some types of plugins, such as equalizers, don't need oversampling at all - but may include it just to make consumers feel better. Limiters,fast compressors, distortion plugins and amp sims are more likely to produce harmonics that exceed Nyquist - illegal frequencies that are incorrectly interpreted and therefore played back at the wrong frequency. That's what aliasing is.

But since this is in the Instruments forum I assume you're talking about oversampling in synthesizers. In that case, oversampling may be called for, but not always. It depends on the oscillator waveform(s) being employed, and the filter settings. The types of patches that might generate audible aliasing are going to be the so-called supersaws, which are extremely rich in harmonic content that can exceed Nyquist. Even then, a well-designed synthesizer will not produce audible aliasing in normal usage.

So if you can't hear the effect of oversampling, don't sweat it. If your ears are good and your monitors are adequate, chances are nobody else will hear it, either. Far more likely is that listeners will create their own nasty aliasing when they play back your music on their smartphones. But you can't do anything about that.

[BlitBit]

It's not cumulative, so even if you have low-level aliasing on many tracks they usually don't add up to audible levels.
...
The types of patches that might generate audible aliasing are going to be the so-called supersaws, which are extremely rich in harmonic content that can exceed Nyquist. Even then, a well-designed synthesizer will not produce audible aliasing in normal usage.

Why should a super saw be prone to aliasing but other waveforms not? After all a super saw is just the sum of some (slightly) detuned saws and if each individual one is band limited and does not show aliasing then the sum of them should not alias as well, should it?

If summing non aliased waveforms could produce audible aliasing this would also contradict the statement about aliasing not being cumulative. Because why should it differ whether the summing is done in a DAW mixer or in the synth oscillator?

But I guess this is indeed something for the DSP forum.

Last but not least, the example that Gamma-UT gave is indeed a nice example that you do not need excellent hearing to notice the effects of aliasing. By the way, if you want another "nice" example of audible aliasing just play some high notes on "Triple Oscillator" in LMMS.