I come from an analog background as an 80s recording musician working in studios and have used software and plugins for 20 years. I'm absolutely no expert with digital audio but I've been noticing some of the comments in posts here about aliasing.

It's never concerned me until now. Now, I've looked into this briefly and not with as much depth and knowledge as some of you have, and I'm a little confused.

Should I be concerned about aliasing? I've never noticed anything to concern me about my digital audio before.

I did my own tests using plugins that have high aliasing and processed a file with them using no oversampling, and again the same file with oversampling. I couldn't hear any difference.

I ran the files side-by-side and inverted the phase of one of them to hear any differences between the two.
There was a noticeable brightness but no apparent junk or distortion I could hear so, what's happening here?

As far as safeguards against aliasing are concerned, what do I do? I read that placing a low pass filer at 20KHz before any plugins will guarantee that there's no artifacts produced by aliasing, and also using dithering - is it that simple to fix?

Obviously, I use as much oversampling as possible to ensure quality, but running every plugin at x16 oversampling isn't resource-friendly and a CPU hog, so what would you suggest?

I typically record projects at 48KHz, 24-bit, using REAPER to track and Cubase for mixing.

Thanks in advance.

There is a lot of tests in this thread: Testing Aliasing of Plugins (measurements)
I think it is depending a lot of what you are doing. Be careful with plugins that tries to emulate harmonics in analog gear. Unfortunately there is no way to guarantee anything. The filter will do the most when it is between plugins that do harmonics.

According to this
https://www.blackghostaudio.com/blog...audio-aliasing

it says that if we are recording at 44KHz or above, aliasing isn't an issue.
The science says that when I'm recording at 48KHz I shouldn't even worry about it - which begs the question - why all the concern on here about plugins with aliasing? Have I misunderstood?

Plugin tries to emulate analog hardware that generate harmonics. This harmonics can be above the nyquist frequency. They need to be handled since they otherwise will fold back and that is aliasing.
Check out Dan Worralls youtube channel. Start with

Thanks guys. Both are very interesting.

Thank you.

If you can hear it *and* it sounds bad, then worry about it.

I am no expert. But from my experience, if you are working in 48k, 24bit projects or higher, and working with sounds that are considered music and musical, you dont need to worry. A lot of these tests are people feeding the plugins sounds that they aren't designed to work with. Its sort of me like saysing, i bet i can really mess up your car engine if i put corn syrup instead of normal gasoline in the gas tank.

I am generally of the "If it sounds good, it is good" camp - but I am really careful with gain these days - I used to think "screw it - we're in digital land - loads of headroom" but many plugins can't deal with hot signals - and when I mean hot I'm only talking -0.5dB - not driving stuff into the red.

Listen to this sweep - it's not extreme 2.5K to 15K at -0.5dB, frequency range and signal strength well within "normal" going through a crappily coded stock Steinberg plug-in (Black) - sounds like a friggin Tardis taking off!

IMHO all those tests for aliasing with extreme settings are useful for finding out the general tendency of a given plugin to create unwanted artifacts, but that does not mean that it is unusable for all signals.
For example, Decapitator will famously alias like hell when you put a sine sweep at high levels through it, nevertheless it is still a perfectly usable processor for say bass guitar or kick drum (even snare or other signals with stronger hi end, if used moderately at appropriate input level).
As with any tool you use, be it analog or digital, ideally you want to be aware of its strengths and weaknesses, the range in with it operates best without creating too many unwanted artifacts and on which kind of signals it works or doesn’t work.
Not always possible, of course, so my personal rule of thumb is: if I expect a process to create higher levels of harmonics (distortion, compression with fast attack/release) and the plugin offers switchable oversampling, I’ll look into that. If it is just a bit of saturation on a signal with mostly low end/lowmids: eh, don’t bother…

Thank you all for the feedback, this is all very helpful.

An other approach is that I can sometimes hear it.
I have to assume that my clients and their clients can do the same. It is reasonable to assume that they might have better ears and gears than me. Therefor there need to be a safety margin.

I remember using a tape sim on the mixbus a modern track I was mixing, being from the era of tape and being used to actual 1970's like noise floors - the slight amount of hiss the sim introduced didn't even register to me (I usually turn it off but forgot.) The young guy whose track it was immediately said "what's that horrible hiss?" followed by "where's the top-end gone?" - what I thought was vibey and warm sounded dull and noisy to the young chap. Conversely, one mix revision later, he loved what I'd consider to be ear-piercing top-end and obvious auto-tune artifacting. Tastes have definitely changed!