KVR Audio

lfm · Post by **lfm** » Wed May 20, 2015 6:09 am

I see this all the time:
- now supports 192k

Another product I saw:
- added support for 384k

What is this about - when is this used?

Are there special up/oversampling algos that need this, or?
Analyzer tools of some sort?

I mean 96k is not uncommon, but still over the top for many considering the cost in performance of a daw etc.

Did you ever use it?

Thanks.

aciddose · Post by **aciddose** » Wed May 20, 2015 6:22 am

The only thing I can think is that they might be used hard coded coefficients rather than an algorithm or equation.

Perhaps there was a bug in the equation they used which caused it to fail above/below a certain rate.

It is also possible the software uses fixed buffers to provide delay effects, for example. In which case higher rates require larger buffers or some limit at which the signal is decimated before being placed into the buffer and interpolated on output.

Ultimately there should be no limit to the range of sample rates you can use other than zero.

Image-Line · Post by **Image-Line** » Wed May 20, 2015 6:27 am

A couple of well written articles on the subject here, listed by increasing complexity:

http://www.trustmeimascientist.com/2013 ... n-it-isnt/

http://people.xiph.org/~xiphmont/demo/neil-young.html

http://lavryengineering.com/pdfs/lavry- ... theory.pdf

Regards Scott

aciddose · Post by **aciddose** » Wed May 20, 2015 7:50 am

When is it used:

When the oversampling or other filter quality in the plugins is lower than desired, it makes sense to increase the rate to the point where the effects of these low quality filters are out of band / ultra-sonic (>20k) and follow the whole thing by a high quality filter+decimation. A single high quality filter will always outperform a series of low quality filters operating at the same rate.
When the effects you depend upon act in the ultra-sonic (>20k) range via intermodulation distortion and this is not handled well by the oversampling filters in the plugins in question. This is very much the same as the first reason, just a minor variation of it.
Much like #2, but when these effects take place between plugins necessitating higher bandwidth to transfer these ultra-sonic signals between the plugins such that oversampling could never achieve the same result. For example imagine a side-chain input with a 30khz signal on it.

lfm · Post by **lfm** » Wed May 20, 2015 8:02 am

Thanks guys - and interesting read Scott.

Effect that benefit should use oversampling - that I discovered too.
Running a synth at 96k(inside Metaplugin) on a 48k project made a noticable improvement in sound.
It probably cover poor internal resampling 44k samples to proper pitch and then to 48k for project.

And converters, I have not fiddled with AD so much, but DAC's - and was amazed on increase in sound quality just changing onboard caps to hi quality ones.

So maybe there is something to my suspision that marketing a lot on 192k or 384k could mean just that design is not that good for 44/48k.

A bit like when a sales person for a car comes with arguments like:
- you know, it has interval wipers

Yes, like that is my biggest concern about cars.

aciddose · Post by **aciddose** » Wed May 20, 2015 11:07 am

I think it's more like talking about wattage on amplifiers, amperes on electric motors or horsepower and acceleration if we're talking about auto dealers.

Yeah sure, it has X horsepower, "do I look like I'm planning on hooking the thing to an 18 wheeler trailer? ...because if I were I'd be more concerned about hauling efficiency to be honest."

(This is the point the dealer's eyes go all glazed over.)

Same with "0 to 60 in X", "do I look like I'm planning on drag racing the thing? ...does it have a roll-bar too?"

lfm · Post by **lfm** » Wed May 20, 2015 12:07 pm

logburner · Post by **logburner** » Wed May 20, 2015 12:46 pm

I use 24bit/192khz setting for digitizing vinyls, and it does make a difference. It was more obvious in 96-192 than 44.1-96 for some reason.

No idea what are the exact factors (frequency above 20hz, alias or else) but tonality of strings and percussive transients sound more open.

When it comes to production/mixing, it's all down to processing power i.e. a powerful enough computer or hardwares. I understand why developers support 192, 384 or even above, because there are studios and people capable of owning and driving such machines. Plus hi-res music is becoming popular. Suppose the future of 44.1 being CD only format is not so distant.

mhog · Post by **mhog** » Wed May 20, 2015 3:11 pm

16 bit 44.1 is more than enough, for me. I can't hear any real difference above this classic standard. When I can, I downgrade 24 bit libraries into 16. This way I can spare in memory and loading time.

kmatheussen · Post by **kmatheussen** » Sat May 23, 2015 1:31 pm

This blog post is quite nice:
http://www.giganticsoftware.com/2015/05 ... ally-hear/

It summarizes, without going too much into the technical details, why there's, from a listeners perspective, no point with more than 40Hz/16 bit. Not only can't you hear the difference, but very little equipment is made to be able to handle more than 40Hz/16 bit, most likely because you can't hear the difference anyway. The blog post also links to a couple of surveys.

When is sample rate 192k and above used - do you use it?