No one can hear frequencies above 20kHz, right? I am not sure what the benefits could be of having not needed high frequencies in your tracks. That's why I am actually wondering if it could be useful or not to apply a high cut with a brickwall filter on it at 20kHz?

Thanks in advance for any help

 

Put analizer on your track, there's nothing above 20k anyway , instruments and effects dont generate above 20k (except some that a purposly designed to)

 

There is anti-aliasing filters already built-in to reduce aliasing/foldover(at all sample frequencies), if that's what you mean.
https://www.ryanschwabe.com/blog/96k

Also, recording in higher SR allows for capturing higher frequencies which is good when doing sound design, etc. I do this quite alot and record in 96k. Pitching down and doing stretching is much more a joy than using "shitty" 44.1 or even 48.

I sometimes high-cut masters over 18k to gain some headroom and reduce some FFT artifacts during coversion to lossy. I learned this from Robert Babicz and it worked for me.

 

Last edited: May 22, 2020

Wrong. Try recording a session in 96k and check the analyzer. It's a bit of an eye opener.

Even at 44.1kHz SR you have the nyquist at 22050Hz. So naturally there is nothing recorded above this frequency, because you can't.
At 96kHz SR you can record frequencies up to 48khz.

 

Last edited: May 22, 2020

If you can't hear it anyways why do you want to put an filter on it?
It's really one of the most ridiculous arguments of those youtube specialists.
Zoom in on this filter then you will see that it's not just a "brickwall". Every filter has artifacts.

 

I have understood that all the frequencies that you can't hear can still generate audible artifacts during the conversion process which could maybe make it handy to just take all these frequencies away that are above it. I mean what's the point of having it if you can't hear it and having the risk of artifacts?

I am not sure though and is why I am asking it

 

Last edited: May 22, 2020

Aliasing means the frequency "bounces back" at the nyquist frequency. It's like flipping the frequency spectrum at the Nyquist frequency. Edited: @Baxter mentioned a good name for that, fold-back aliasing.
So if you have noise at 28Khz it'll become 20Khz aliasing when recording using 48Khz (24Khz Nyquist). And it can't be removed because it's mixed with the original 20Khz stuff.

Not that I think it matters but theoretically could be there.
This could be easily resolved using a 20K low pass filter at the hardware level, no matter the sampling rate you use after that.
So if want to look professional just say you record at 192Khz and then filter plus resampling. They can't tell the difference anyway.

10 years from now it'll be 768K probably, if only for the snob-elitism factor we all know and love.

 

Last edited: May 22, 2020

Aswell as all the freqs you can hear.

 

Last edited: May 22, 2020

usually doing stuff just 'because' is bad.

 

additionally. during mastering, it isn't uncommon to use high shelving filters above 22050Hz with low Q factor
so it still affects the audible range, even on projects delivered in 44.1.

The problem is with inserting a highcut brick wall just 'because' is that the phase shift is going stack with whatever else is going on later on,
will this produce artifacts? probraly not, the later adjustment doesn't make the cut.

 

Not really. The problem are the frequencies ***above the nyquist limit*** that are being generated and can't be heard originally and end up reflecting back into the audible spectrum after going through a dynamics/saturation plugin. It's what is called aliasing and it sounds pretty much FM-ish [like FM synthesis].

If you cut those high frequencies out before doing any *digital* dynamic or saturation processing, you will get a much cleaner sounding result.

To explain this further, try doing this: generate a high frequency sinewave you probably can't hear [~18kHz] and put some saturation plugin on it. Play with it, if you can, for even better effect. You will be able to hear the FM-ish aliasing easily. Now put a LPF [no resonance] before the saturation plugin and have a little play with it downwards. You'll hear the aliasing going away the more high frequencies you cut before entering saturation and understand the Nyquist theorem completely.

Cheers!

 

Last edited: May 22, 2020

I wasn't referring to the Nyquist problem, but to the lossy 128kbps conversion. This can also affect freqs way below Nyquist.

Err, which cutted high freqs when I use a sinewave???

 

The sinewave isn't the best example here...

 

+1 on cutting 18khz and above from @Baxter. Helps with headroom and interfering high end...

 

In 10 years I demand infinite SR for my quantum CPU, my qWatch, qPad and qPhone.

And of course for qBase.

 

Guys, what headroom? The level of the highest freqs can easily be two-digits below the lowest freqs and tbh, they should be.

 

Fabfilter has closed this topic with an amazing video.

PS. Personally I think that higher samplerate is actually better, except if all of the VSTIs and harmonic freq generating plugins you use have built-in oversampling.

 

"The world should reach 8.5 billion people by 2030, up from 7.3 billion in 2015. The fastest growing demographic will be the elderly, with the population of people over 65 years old at 1 billion by 2030".

At least One billion people that can't hear over 12.5/14khz. Will high freq filtering make a difference for Gen STBPUD (soon to be pushing up daisys)?

 

Last edited: May 22, 2020

WHAT DID YA SAY? CAN'T HEAR SH*T YOUNGSTER!!

Not a thing. But someone has to fight against the TVC generation (The Vegan Conspiracy)

 

Well, I think a very high sample rate of like 192 for instance is not going to give you an audible better audio quality. Well, in theory it could be and likely is, but it's also important that people can actually hear it. Instead I think this is just a form of marketing to seduce people to still buy new audio interfaces and so they can continue with earning money in the future. I don't think people can hear any differences when using very high sample rates though

By the way I am not sure why the limiter from Fabfilter is having oversampling of x32 while they are saying at the same time that higher oversampling can generate ringing effects.