everything above 22k cant be heard anyway so who cares?

 

Higher sample rates (like 96kHz, 192kHz) are not for hearing everything above 20k and even not for preserving these frequencies (in my case and many engineers).

Learn more!

 

Drawbacks of High Sample Rates

As mentioned above, the human ear can only hear certain frequencies. In theory, a higher sample rate will only capture frequencies at extremely high and low ends of the spectrum where listeners can’t even hear them. This means you’re spending more and using more space for music that doesn’t have a noticeable improvement in sound.

... this is my point... some normie walking around in beats headphones that actually buys music doesnt give two shits about any of this... so ultimately who cares other than the engineer

 

96khz+ isn't for preserving anything (you can't hear it bro, even if its there), the only reason to use anything higher than 44.1 is either to lessen aliasing (if you use plugins that dont have internal OS) or to comply with video. Apart from that there isn't a single practical reason to work with higher sample rates.

 

Again! You are thinking about higher sample rates only as "hearing or not hearing" ultrasonic stuff. But these higher sample rates were introduced not for that! Yeah, maybe DAC outputting higher sample rate audio as bit more detailed, less distorted, not too filtered at circa 20 kHz area. But when you are working at higher sample rates, or oversample lower sample rates - it is the another story.
Search, learn. There is a lot of hints given here or around this thread.

Hint again: higher sample rate is for processing,... And there are minimum 3 cases why you need such higher sample rates

 

3 reasons (5 reasons sorry) :

1. Lessen/minimize/remove aliasing
2. Equing highs at nyquist area (both frequency, phase)
3. Pitching

And yeah... Video formats.... (4th reason)

5. Dithering and noise shaping for hi-res stuff.... Noise shaping at 44 still will be louder for your ears than noise shaping at 48.... Higher sample rate, a la at 192kHz, is almost non-hearable =)
Depends on algos of all dithering/noise-shaping utilities working at different sample rates.... But you may check Mbit+ dithering algo in RX

 

Last edited: Feb 10, 2022

Also, an interesting fact...
A high-cut filtered audio (above 20k) or upsampled audio (from 44.1k to 96k, for example) will be better for saturators/distortion units, than full spectrum with ultrasonic stuff.
Why? There will be less aliasing, because we have (almost) nothing above 20k. Aliasing just will not appear from there.
"Upsampling will not improve audio" is a myth. Upsampling will lessen aliasing from useless ultrasound. Same with filtering.

 

yeah, thats basically what I said, processing and video compliance. but even then anything above 48.000 is overkill imo.

edit: sorry, I thought I said processing but I just mentioned aliasing, fair enough, good point

 

Keep in mind that the result is cumulative. Have 32 or more tracks oversampled using interpolation or low-q sinc64 [I put it to sinc192 in Reaper but don't use it], render the project, and compare to the same project rendered using Voxengo r8brain oversampling.

Not saying that you might like the oversampling artefacts you can hear using low-q oversampling. These days music is "mastered" so badly that everything sounds like flat mush anyway and most people don't care, or think that's the way it should sound etc.

There is a very nice website https://src.infinitewave.ca/ where you can compare most if the sample rate conversion algorithms. Try comparing Reaper [HQ-means sinc768] with Voxengo 1.9 free SRC. Then take a look at Native Instruments Kontakt 5.71 and try not to vomit. That's why it's rather important to use kontakt sample banks and samples at the same sample rate they had been recorded at [or just render Kontakt track at the sample banks sample rate and resample with r8Brain to the sample rate you work with as I do ;)].

Cheers!

 

There's VladG's simple plugin that does exactly that - nicely filters out frequencies above 20kHz or thereabouts. Great plugin, using almost no CPU.

@madbuzzin You can't hear anything above even 18kHz, probably even lower, however, when you use SRC the frequences bounce back from the Nyquist limit into the audible range we can absolutely hear. Go to the https://src.infinitewave.ca/ and take a look at the SRC test graphs. Especially comparison between Kontakt and r8Brain, so you'll [probably] understand what I mean. The cleaner looking graph, the better.

 

Last edited: Feb 12, 2022

I know) TDR Ultrasonic filter and several versions of such filters by Airwindows.

Yes, I know it too about infinitewave. I did some test and they posted it there. I did Aimp conversions, Nero, dbpoweramp 15th, finalcd goldilock filter and couple of others.
The most fine are by SoX (free command-line, Ableton Live has it since v9.11 or so), voxengo r8brain 2 (pro/free, recently appeared in Reaper), finalcd sharp mode, RX (fully customizable, freeware, most cleanest and sharpest), dbpoweramp (most cleanest and sharpest, same as finalcd) weiss saracon (very good, not ultra top), the brick (mac only, freeware, abandonware, very clean and sharpest), some others.
Foobar2000 introduced src, same one as in dbpoweramp, right after my tests on infinitewave. It was discussed on their forum, requests done and voila, it is in foobar too etc.

 

When I'm using Windows, I use Foobar2000 for a lot of things. Such a useful program with excellent SRC plugins. Pretty much not very different from Voxengo r8brain. I miss it in Linux. However, now I have r8brain in Reaper v6.47.