Well, thanks for your contribution, but if you want to pitch your recorded sound material by several octaves, oversampling won't help you. Oversampling also doesn't help you with processes where oversampling is not an available option or is implemented suboptimally. There are also sonic differences between oversampling and higher project or device sample rates. There are even sonic differences between different oversampling implementations. Using a process in a wrapper is an additional option, but it costs additional steps - and thus additional time. Disk space and powerful CPUs are no longer overly expensive resources. Especially if you can deduct them from your taxes. What do i care about 5-6% more CPU load for a medium sized project, if there is still more than 60% available? It's all about the cost-benefit ratios. For me, this approach pays off several times over. If high sample rates cause you more problems than benefits, for example because your equipment can't keep up, fine. No one tells you what's best for you. You can only answer that for yourself. And that is good! Everyone has different needs and demands. And I just need high sample rates. And sometimes the customer does too.

 

IMD has nothing to do with the sample rate. A high sample rate also does not introduce or increase IMD. IMD is a product of harmonics in nonlinear systems and is strongly dependent on the level of the fundamental. This is true for all nonlinear systems: your microphone, your preamp, other outboard processors, your speakers.... But if you don't generate high level harmonics in the high frequency spectrum like a completely insane maniac, you won't have a problem with IMD. Even if you work with 768kHz.

What do all these engineers do who work with their analog equipment whose filters limit the spectrum at 80-100kHz? All these great mixes of the last 60 years... completely ruined by IMD. NOT!


EDIT:
I think what Dan was trying to explain is that a high sample rate is not a problem solver for every problem. When you work with a high sample rate, you are working with a frequency spectrum that is out of your hearing range. This means that problems may occur whose cause is not immediately apparent because it is outside your hearing range. And only if you are aware of this fact, you can solve the problem.

Just by the way, so you don't misunderstand me: IMD also occurs when you are working in 44.1kHz. More precisely, if you don't happen to make music with single sine tones, you will get IMD as soon as you use non-linear processes. Always. And there is absolutely nothing you can do about it. IMD is one of the reasons why we use saturation. The question is, does IMD cause a problem in your mix? If so, don't blame a high sample rate, blame the source: the frequency contour of your instrument or your nonlinear processor. Especially with an asymmetric distortion curve, IMD is something to keep an eye on... And then blame yourself! Learn and understand additive synthesis. And then solve the problem.

 

Last edited: May 30, 2022

This is all true, but your argument is on a purely semantic level. This entire thread is based on the question of whether 192kHz is useful for recording/production/mixing, and if so, what is the benefit. Both questions have been answered. Answer to question 1 is yes. And the answer to question 2 is ... a bit more complex. But no one has asked if you can hear the bandwidth outside the audible spectrum. Because the question answers itself. And, as far as I can tell, no one has claimed that a high sample rate brings any advantages in reproduction. Well, not quite right... I have claimed that. But my speakers have an internal DA, so the situation is different.

Working at a 48kHz project sample rate with digital nonlinear processors and 16x oversampling is about the same as working with analog nonlinear processors whose filters in the output stage are at about 24kHz.

Working at a 192kHz project sample rate with digital nonlinear processors and 16x oversampling, on the other hand, is roughly equivalent to working with analog nonlinear processors with filters in the output stage at about 96kHz.

Even if we leave oversampling completely aside, both methods can lead to completely different sonic results, since sounds generated outside the audible spectrum can interact with each other and influence the audible spectrum. Especially when non-linear processors are connected in series. In the first method, however, there are no harmonics above 24kHz that can be modulated. Neither is generally good or bad (because what is good or bad is decided by the intended use and taste), but the resulting difference is both audible and measurable. And since you don't miss the opportunity to constantly refer to the AES:
AES has more than adequately studied these phenomena. Both - inside the box and in analog signal chains.
So what are we actually talking about here? Semantics?

 

Last edited: May 30, 2022

Although, I didn't want to reply here any more...
My replies were specificly towards two people who claimed different things.
Obviously, one can feel and hear the differences created due to Oversampling because without OS, a lot of higher harmonics do fold back to audible spectrum creating components that are,
A. Not at all related to the Music and
B. Doesn't change the same way the actual fundamental does, as in Fundamental going upward while IMD and Foldbacks going downwards and vice versa.

Similarly, there are a good Time stretching Quality aspect related to higher sample rate.
But in both cases, you can't feel the difference of 192kHz to 48kHz.
You feel the artifacts of a second processing.

Also, a guy talked about recording Guitars etc and listening to it.
Again, that's a biased experience.
Why?
When you record a Gtr, at whatever sample rate, the antialiasing filter will always work at the highest sample rate of the device.
Say you have a good ADC which supports 192kHz.
So if you record a Gtr at 48kHz, it will be antialiased at 192kHz and then it will be recorded at 48kHz.
Similarly if you record the same Gtr at 44.1kHz or 96Khz, it will also be antialiased at 192kHz and then will be recorded at 96kHz.
So in both cases, unless you add plugin or processors (Hardware) to create Non Linear processing, you will not be able to hear the difference in blind tests.
You will hear a difference without blind test as a basic result of Confirmation Bias.
That's what I was trying to say all along.

As per the answer to the original thread.

If you are going to do a lot of time stretching based processing, 192kHz can be beneficial not because you hear a difference but because the product after radical time stretching will be of better quality.

For mixing, as per many tests (I have done this using DDMF plugin doctor as well), it is better to have plugin level Oversampling than overall higher sample rates.
This has been mentioned here as well.
Plus, when you use oversampling on per plugin level, the plugin coding, over sample only certain parts.
Remember, a good Look ahead and many other features are dependent on Filters and filter quality is very much dependent on sampling rate.
So if you have systemwide higjer sample rates, chances are those filters will be fucked.
This is why it is better to use plugin level oversampling.

This is even true for the new Reaper feature where you can oversample an entire chain of plugins. That can be highly dangerous resulting in complete fuck up in some cases and so it is way better to use plugins that allow oversampling natively because then only certain parts (eg Sidechain) will only be oversampled and not the entire audio and the entire processing.

Also, you, unlike others have written some data driver points without being a jerk and so I am replying to you thoroughly as per what I know or tested wih whatever devices and plugins I have.

I see you talked about IMD.
(The interact with each other part).

And honestly, IMD may or may not be heard depending on the level of it and a lot of psychoacoustic phenomena that helps us mask a lot of lower level details.
(This was the basis of the data reduced perceptual encoding format)

I love to read AES journals and papers and it really gives me so much joy to read those because, I would like to start doing some sort of audio related research someday.

 

you can t hear it, and i excplained you why, go to have your ears checked

 

do it with the voice,probably you can t do that either and prefer to remain stuck to your fantasy world

guess what, if the quality of the product is better you will hear a difference.
Seriously, stop it

 

Make way for the tough guys