192 kHz in real life ... quite surprising ...

[ISedlacek]

I have quietly lived within an idea that 192 kHz is something beyond the hearing spectrum, good just for bats and marketing people.

About a week ago, I got a private message from another GS member where (apart from other things) he mentioned his surprising experience when comparing a choir recording done in 96k vs 192k ...

I was a bit sceptical, but the best thing is to try ... I recorded few things in 192k, the same in 96k , converted 192 k samples to 96 and compared ... well ... they did not sound the same... But I did not want to come to some hasty conclusions and kept playing with it for the next few days ... and the results were the same (whatever instruments I played).

The first thing that was quite different and noticeable was the depth - 96 k samples sounded thinner as if comparing to 192k samples. Another factor was the HF content: in 96k samples, slightly "abrasive" digital "brush" was clearly present, while (former) 192k samples had quite a round and full rendering of this area, more "analogue" so to say ... I talked to Mikael Vest from DAD and he said, that the basic difference may not be caused by the sample rate itself, but by the filters ... that work somehow less obvious in higher sample rates (while in 384kHz they are no filters at all)

The difference between recording in 44 vs 96 is already big. 96 vs 192 may not be exactly that big, but quite big ... after hearing the differences so many times and in so many cases, I think my next projects of my subtle acoustic music will be starting in 192 kHz ...

One lives and learns ...

[mohthom]

And which hard drive manufacturer will be sponsoring this noble endeavour?



MohThoM

[ISedlacek]

Originally Posted by mohthom

And which hard drive manufacturer will be sponsoring this noble endeavour?



MohThoM

Oh, these HDDs cost almost nothing nowadays, I would not worry about that

[hummer]

It is true that lower order filters are used at higher sampling rates, but the adverse effects of filtering are eliminated after a Nyquist frequency of about 30 kHz (regardless of filter design), which is a 60 kHz sample rate. There is plenty-o-research to support that...

Additionally, sampling at 192 will typically yield a less precise result in terms of amplitude quantization. (higher rate of sampling = less bits = more quantization distortion) From my understanding, most converters operate optimally at around 96kHz.

There are definitely grounds for supporting the use of higher sampling rates/additional bandwidth when a significant amount of signal processing is going to be used, particularly time compression/expansion...

I'm not saying that you don't hear a difference, I'm saying that it may not be due to the reasons you think.

Remember Nyquist/Shannon: All of the audio information is contained in the sampled signal, provided that the sampling rate exceeds twice the highest frequency in the sound being sampled. --- Unless the microphone you're using is outputting signals above the 96k Nyquist frequency of 48 k (very few pick up anything beyond our hearing threshold) then it's safe to say that you're hearing hardware influence of some sort.

Speaking of which, what hardware are you using?

[hummer]

Oh and also, you said that you SRC'd the 192 recording to 96 BEFORE you compared? Why would you do that? You're also hearing the influence of whichever SRC algo you used.

[zoom]

...and this was a double blind test??

DBT - The only way for really serious investigations in that matter.

[ISedlacek]

Originally Posted by hummer

Speaking of which, what hardware are you using?

I use Forssell MADC-2 converter. I also wanted to use DA AX-24, but somehow still did not make it work for 192 recording (while 192k DA works fine).

Originally Posted by hummer

Oh and also, you said that you SRC'd the 192 recording to 96 BEFORE you compared? Why would you do that? You're also hearing the influence of whichever SRC algo you used.

Sure, but first - having two files of a different sample rate in one session is not possible. Second, the point is how the result sounds when you bring it back .. (normally to 44k). There is no use if it sounds great in 192 but after conversion to 44k it sounds the same as if originally recorded in 44

[klaukholm]

I had a nice talk with Michael about this at their office last month.


His filter argument is a strong one and I will have to give his ADs a go when I have the time and the right project.
I go the impresion from him that the difference is bigger from 192 to DXD than 96 to 192.
This was not really marketing talk as he is well aware that we are up to our teeth in AD's with no intent to purchase for a long while.

[Ozpeter]

Unless the most rigorous testing is undertaken (ABX, and quite possibly head in a clamp!), using monitors that can be shown to actually reproduce the frequencies of interest, such considerations come down to matters of individual taste and personal preference - there is no such thing as the best wine!

Of course we have to do the best we reasonably can as professionals, but certainly there is not a shred of evidence to connect high sample rates with the buying preferences of the public when it comes to the distributed recording. There's an increase in cost of production but no increase in sales of the finished product - not good business (except for those selling the gear).

[Chris Lago]

Regular people won't hear the difference, so I don't see what the point is really.

[NesNeedsGear]

Originally Posted by chrislago

Regular people won't hear the difference, so I don't see what the point is really.

yep

not too many people can afford converters/monitors/room good enough to notice the difference.

do people even care??

[desotoslo]

Ivo cares. nuff said.

[DSD_Mastering]

Originally Posted by desotoslo

Ivo cares. nuff said.

I care as well.. as do the thousands of people downloading our files at HDtracks.com


Regards,

[desotoslo]

cool thumbsup I just meant, it's up to you [and your clients].

[aracu]

In comparing sample rates I suggest creating blind tests for yourself to see
how much of the difference is caused by psychological factors. It is much less
time consuming to work with lower sample rates.

[DSD_Mastering]

Originally Posted by aracu

It is much less
time consuming to work with lower sample rates.

Working with higher sampling rates is no more time consuming than working with lower rates if you have the proper equipment. It's just a matter of pressing a different button.. that's it.

Regards,

[Ozpeter]

Working with higher sampling rates is no more time consuming than working with lower rates if you have the proper equipment.

Kind of depends what you are doing. For instance, given that there's a lot more data to be handled, when doing non-real-time operations (eg rendering a finished recording in your DAW) there's bound to be an increase in time taken. Whether in the overall scheme of things that matters is another thing.

[ISedlacek]

Originally Posted by chrislago

Regular people won't hear the difference, so I don't see what the point is really.

Regular people won't most probably hear the difference between RME and Lavry Gold, between Mackie preamps and Neve, between Beyer and U47 etc etc. in the final result. You can easily save lot of money ... and keep cheap factory production. But are we always doing all this for "regular people" (with a vision of their ipods and computer speakers), or sometimes also for the sake of beauty and our own pleasure, joy and satisfaction ?

The only client I record the music for, is quite demanding one and really wants choosing and comparing the best sounding options, capturing the utmost depth and most subtle details of the music. He does not pay much though. His name is Ivo

Originally Posted by Ozpeter

Unless the most rigorous testing is undertaken (ABX, and quite possibly head in a clamp!), using monitors that can be shown to actually reproduce the frequencies of interest, such considerations come down to matters of individual taste and personal preference - there is no such thing as the best wine!

Of course we have to do the best we reasonably can as professionals, but certainly there is not a shred of evidence to connect high sample rates with the buying preferences of the public when it comes to the distributed recording. There's an increase in cost of production but no increase in sales of the finished product - not good business (except for those selling the gear).

Sometimes there is clearly one wine better than the other ... But we are not talking of any business and economy ... I am not buying a great microphone, preamp, converter, compressor, EQ etc. with the clear vision how much money it will bring to me (it will not). I have different reasons for it, I am not a factory ...

But I did not bring this topic in order to persuade anyone that 192k sounds better. I just shared my own experience (and I have no affiliation whatsoever with the higher sample rates producing company ) and the only way to get something out of it is to try for oneself and see (or not , if not interested, but then probably not knowing how it really works). But the most remarkable (and quite expected) thing is the tendency to find 1000+1 reasons why 192 (or whatever) simply cannot be better/regular people will not hear/will be complicated, time consuming/expensive/deluding without 108 blind tests etc etc - without actually trying it ever for oneself - which is actually quite easy. Maybe there will be different results with different converters, maybe different people will find different things, but without it, we return to the well known tale about the fox and grapes )

I could also have posted same AB samples, but to say the truth I am already a bit discouraged by some anticipated comments like : " I have just quickly listened on my laptop speakers and found not much difference" and besides, the file B is 0,00000004 dB louder that makes the whole comparison non-valid. (and the others will just often take it as a relieving fact without any further listening)

I will for sure experiment and compare more in depth trying to get to some definite conclusion

[fondone]

I want dan lavry to prove to IVO that 96khz is all he'll ever need!

this is one of the greatest misconceptions of digital audio.....

dan lavry thinks 96khz is theoretically enough......


Is it true IVO?

[ISedlacek]

Originally Posted by fondone

I want dan lavry to prove to IVO that 96khz is all he'll ever need!

this is one of the greatest misconceptions of digital audio.....

dan lavry thinks 96khz is theoretically enough......


Is it true IVO?

I am not recording theory, but music ... and rather than reading about it, I want to listen to it ... The new Lavry Black DA 11 has 192k sample rate by the way )

Whether it is sample rate, or filters or whatever that causes the sound difference is secondary to me ....

I remember a similar discussion long time ago at another well known forum about 44 vs 96k recordings. I posted some samples where the sound differences were clearly almost night and day. Here on GS many people heard them clearly. But that was very theoretical subforum (that does not exist any more) where the members had very precise theoretical idea that 96 k simply CANNOT sound better than 44k, there is not a single theoretical reason for it. And even if it sounds, it does not Of course, they could not say that the AB samples sound the same, because they obviously did not ... Then they said that the reason is the performances were not identical and when recording in 96k I somehow wished it sounds better and therefore played better Then I removed dust from my old Roland XP 50 and recorded few loops in 44k and 96k (where the "performance" was clearly impartial). Then they had to admit there IS a difference (at least those who cared to listen), although it has no theoretical reasons. So they came to a sudden idea that this is because of filters ... well, does not matter for me, as long as it sounds better ...

[mathieujm]

Originally Posted by hummer

Additionally, sampling at 192 will typically yield a less precise result in terms of amplitude quantization. (higher rate of sampling = less bits = more quantization distortion) From my understanding, most converters operate optimally at around 96kHz.

I don't understand that. Don't the 192k converters have a 24 bits quantization ?

JMM

[bcgood]

192 sounds good 5.6 MHz DSD sounds even better.

[ISedlacek]

Originally Posted by bcgood

192 sounds good 5.6 MHz DSD sounds even better.

This is not unlikely, but already out of common reach at the moment (no normal DAW renders higher than 192 k rates now - except Pyramix etc)

[Audiop]

Originally Posted by mathieujm

I don't understand that. Don't the 192k converters have a 24 bits quantization ?

JMM

Yes, and the QD is a moot point as I see it since it is turned into noise at a level way below the analog noise of the converter.

Modern converters perform almost equally well at 44.1kS/s and 192kS/s. The difference beeing potentially less aliasing due to more freedom with filter design at higher sampling rares. Also if the pre ringing in linear phase digital brickwall filters is audible it is a good reason for higher sampling rates.

I record at 192kS/s because my converter performs audibly better. Still must set up some tests to see where the "breaking point" is. IOW if 88.2kS/s or 96kS/s is indisinguishable from 192kS/s or not.

My simple set up with a "outdated" 12" laptop have no problems capturing several hours of 8 channels @ 24/192 into the internal HD.


/Peter

[Audiop]

Originally Posted by ISedlacek

I am not recording theory, but music ... and rather than reading about it, I want to listen to it ... The new Lavry Black DA 11 has 192k sample rate by the way )

And there goes his credibility..


/Peter

[Audiop]

Originally Posted by bcgood

192 sounds good 5.6 MHz DSD sounds even better.

Subjectively or "absolute"? IOW investigated in controlled listening tets or experienced in day to day work?


/Peter

[bcgood]

Originally Posted by Audiop

Subjectively or "absolute"? IOW investigated in controlled listening tets or experienced in day to day work?


/Peter

Absolute

[Audiop]

Thanks!

Would you mind telling some about the test and gear?


/Peter

[bcgood]

Originally Posted by Audiop

Thanks!

Would you mind telling some about the test and gear?


/Peter

I don't know, it's pretty technical stuff. I'm not sure you would understand everything.

[HDJK]

Originally Posted by ISedlacek

The new Lavry Black DA 11 has 192k sample rate by the way )
...

Originally Posted by Audiop

And there goes his credibility..


/Peter

It's a DA. People record at 192 and he provides a way to convert it. No 192 AD in his line-up, credibility saved ;-)