COCKOS
CONFEDERATED FORUMS
Cockos : REAPER : NINJAM : Forums
Forum Home : Register : FAQ : Members List : Search :
Old 06-30-2016, 05:34 AM   #1
Fabian
Human being with feelings
 
Fabian's Avatar
 
Join Date: Sep 2008
Location: Sweden
Posts: 7,416
Default I read the news today...

"The overall conclusion is that the perceived fidelity of an audio recording and playback chain can be affected by operating beyond conventional resolution."

http://www.aes.org/e-lib/browse.cfm?elib=18296
__________________
// MVHMF
I never always did the right thing, but all I did wasn't wrong...
Fabian is offline  
Old 06-30-2016, 09:09 AM   #2
serr
Human being with feelings
 
Join Date: Sep 2010
Posts: 12,557
Default

This has been a done deal for me for a long time now. Modern HD digital (eg. 24 bit 96KHz) captures full audio far down into the decimal dust. You'd need an insane level of 'money no object' analog gear to almost equal it. You can't blame any audio issues on the recording format anymore, nor the consumer delivery because it's the exact same file format direct to the consumer.

But right away in the 2nd sentence:
"...high resolution and standard (16 bit, 44.1 or 48 kHz) audio."

Equating sample bit depth with sample rate?!
Such a fundamental misunderstanding of the subject isn't such a good way to start an article. If that's sloppy editing... well, that doesn't cut it either.

This is certain to degrade into the usual claims of low bit rate mp3's being the highest quality recording format known to super human hearing and physics with all the strawman arguments to back that up.
serr is offline  
Old 06-30-2016, 01:20 PM   #3
Fabian
Human being with feelings
 
Fabian's Avatar
 
Join Date: Sep 2008
Location: Sweden
Posts: 7,416
Default

Quote:
Originally Posted by serr View Post
But right away in the 2nd sentence:
"...high resolution and standard (16 bit, 44.1 or 48 kHz) audio."

Equating sample bit depth with sample rate?!
Such a fundamental misunderstanding of the subject isn't such a good way to start an article. If that's sloppy editing... well, that doesn't cut it either.
I think you misunderstand that (if I understand you correctly). They call 44.1/16 and 48/16 "standard", while "high resolution" is typically (but not in all studies) 192/24.
__________________
// MVHMF
I never always did the right thing, but all I did wasn't wrong...
Fabian is offline  
Old 06-30-2016, 01:23 PM   #4
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 29,260
Default

Quote:
This is certain to degrade into the usual claims of low bit rate mp3's being the highest quality recording format known to super human hearing and physics with all the strawman arguments to back that up.
As far as 320kbs, I think the only argument is for those saying they hear it to pass a blind ABX test. I'm not stating or have an opinion either way, just saying that anyone taking a stance that they can hear it should step up to the plate and prove it.
__________________
Music is what feelings sound like.
karbomusic is offline  
Old 07-22-2016, 06:40 AM   #5
noitall
Human being with feelings
 
Join Date: Apr 2014
Posts: 245
Default

....."Oh boy".....Sorry, couldn't resist.
noitall is offline  
Old 07-22-2016, 02:49 PM   #6
Dannii
Human being with feelings
 
Dannii's Avatar
 
Join Date: Mar 2010
Location: Adelaide, South Australia (originally from Geelong)
Posts: 5,598
Default

Quote:
Originally Posted by noitall View Post
....."Oh boy".....Sorry, couldn't resist.
Beat me to it!!! That was my first thought when I saw that title!!
__________________
Dannii is offline  
Old 07-22-2016, 03:20 PM   #7
alanofoz
Human being with feelings
 
alanofoz's Avatar
 
Join Date: Sep 2009
Location: Oz - Blue Mountains NSW, formerly Geelong
Posts: 943
Default

Quote:
Originally Posted by noitall View Post
....."Oh boy".....Sorry, couldn't resist.
Quote:
Originally Posted by ReaDave View Post
Beat me to it!!! That was my first thought when I saw that title!!
Yep, very appropriate in more ways than one.

There's a long discussion about this here. The first page is probably enough, beware of some of the opinions.
__________________
It's "its" except when it's "it is".

alanofoz, aka Alan of Australia
alanofoz is offline  
Old 07-22-2016, 04:02 PM   #8
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Thanks for posting that. (Hopefully a mod can move this thread out of the spam trap...)

So, just to play the curmudgeon:

They didn't (unless i missed it) disqualify any studies for financial bias. E.g. citation 70, where high-res audio was discriminated, was apparently funded by a company developing a high-res car audio distribution system (judging by the Acknowledgements section). Citation 44 is described by wikipedia as a "controversial" paper, and the author of that study himself apparently designed and/or sold hypersonic tweeters. Citation #23 found discriminatory ability, and was funded by a loudspeaker company that makes speakers that produce frequencies up to 40kHz.

I haven't gone through all the citations (those are literally just the first three I checked). And I'm not trying to slander the author; maybe big-tent inclusion is par for the course with such meta-studies. But it makes me a little more hesitant about the results. I want to see an additional statistical breakdown based on papers that weren't funded by companies that develop or promote products that stand to gain from such discriminatory ability.

It's also worth pointing out that even as it stands the conclusion isn't all that strong: "Overall, there was a small but statistically significant ability to discriminate between standard quality audio (44.1 or 48 kHz, 16 bit) and high resolution audio (beyond standard quality). When subjects were trained, the ability to discriminate was far more significant. The analysis also suggested that careful selection of stimuli, including their duration, may play an important role in the ability to discriminate between high resolution and standard resolution audio." -- it doesn't say anything about higher-res sounding better or clearer or more accurate, etc., just that discrimination is possible.

This is certainly an interesting paper... i'm not qualified to evaluate the very involved statistical wrangling they did to try to bring all the data together, but it seems like earnest work and something to be reckoned with.

I look forward to reactions/criticism by qualified folks. I'm mainly glad that people are trying to take a serious look at the issue.

Last edited by clepsydrae; 07-22-2016 at 05:13 PM.
clepsydrae is offline  
Old 07-22-2016, 05:06 PM   #9
grinder
Human being with feelings
 
grinder's Avatar
 
Join Date: Jan 2009
Location: New Zealand
Posts: 2,905
Default

Good to see this
I can hear the difference between 44 and 96 quite clearly.
It must be said I have been in the studio quite a while and listen mostly at low levels, live in the country, can still hear 16Khz and am 64 years of age.
I wonder though how many people especially those who live with the noise of constant heavy machinery, workplaces, ultra loud music and city life still have quality hearing?


Grinder
grinder is offline  
Old 07-22-2016, 06:35 PM   #10
tack
Human being with feelings
 
tack's Avatar
 
Join Date: Jan 2014
Location: Ontario, Canada
Posts: 1,618
Default

Quote:
Originally Posted by grinder View Post
I can hear the difference between 44 and 96 quite clearly.
What about 48 and 96? Did you test your gear for intermodulation distortion? Any gear that fails IMD tests can't be used to draw any conclusions about this. (And you'd be surprised at some quality gear that has IMD problems.)
tack is offline  
Old 07-22-2016, 06:36 PM   #11
tack
Human being with feelings
 
tack's Avatar
 
Join Date: Jan 2014
Location: Ontario, Canada
Posts: 1,618
Default

BTW Ian Shepherd talks about this paper in an episode of The Mastering Show.

http://themasteringshow.com/episode-17/
tack is offline  
Old 07-22-2016, 07:02 PM   #12
grinder
Human being with feelings
 
grinder's Avatar
 
Join Date: Jan 2009
Location: New Zealand
Posts: 2,905
Default

Quote:
Originally Posted by tack View Post
What about 48 and 96? Did you test your gear for intermodulation distortion? Any gear that fails IMD tests can't be used to draw any conclusions about this. (And you'd be surprised at some quality gear that has IMD problems.)
I guess there must be heaps of avenues to influence results!
I have recorded at 48 to me it sounded a bit smoother than 44.1
When I listen to 96 compared to the smoother sound I hear at 48 from 44.1
96 turns to be much smoother and richer, like opening a door to the outside. I came on this before I read about any comments regarding what sample rate was the best and the arguments that ensured.
I use an Orpheus.

Grinder
grinder is offline  
Old 07-23-2016, 08:00 AM   #13
tack
Human being with feelings
 
tack's Avatar
 
Join Date: Jan 2014
Location: Ontario, Canada
Posts: 1,618
Default

Quote:
Originally Posted by grinder View Post
96 turns to be much smoother and richer, like opening a door to the outside.
There are many reasons why one should be skeptical of statements like that:

1) They often collapse under properly blinded testing.

2) The testing itself is often done incorrectly. One example: in comparing 96kHz vs 48kHz, one must take 96kHz content, downsample to 48kHz, and then (losslessly) upsample back to 96kHz for playback. This is because DACs can have entirely different filters for 48kHz vs 96kHz, and you may not be testing your ability to perceive higher frequency content but instead testing the quality of your particular DAC's filters.

3) The signal path is often not verified to be free of IMD. You should try these tests. My respectable RME Babyface suffers from IMD, for example. This generates distortion in the audible spectrum which can actually sound pleasing. https://xiph.org/~xiphmont/demo/neil-young.html#toc_1ch

The meta-analysis isn't without its problems as a result of not having consistent controls and testing methodology handling the issues mentioned above (or at least 2 and 3, since IIRC they removed all non-blinded data from the study set). Taking into account the failure to properly control for these real-world problems across the set of studies analyzed, the small effect size observed in the meta-analysis is really underwhelming. (The meta-analysis adjusted the weighting of studies that didn't control for IMD, rather than completely eliminating them. I find this spurious because that alone could account for the small effect size.)

Last edited by tack; 07-23-2016 at 08:36 AM.
tack is offline  
Old 07-23-2016, 10:49 AM   #14
toddhisattva
Human being with feelings
 
toddhisattva's Avatar
 
Join Date: Jun 2008
Location: Austin
Posts: 289
Default Metan Alysis

Oh, a meta-analysis. That's all I need to know. Too lazy to conduct their own experiments.
toddhisattva is offline  
Old 07-23-2016, 10:56 AM   #15
tack
Human being with feelings
 
tack's Avatar
 
Join Date: Jan 2014
Location: Ontario, Canada
Posts: 1,618
Default

I think that's unfair. Meta-analyses have the benefit of much wider data sets than single studies. Obviously a single study with an equal amount of data would be much better, but that's almost never practical. It's not a matter of laziness, but of funding.

The trouble is that meta-analyses are obviously problematic because they try to unify often substantially different studies with entirely different biases. Also meta-analyses tend to be more susceptible to p-hacking.

So, yeah, for that reason a meta-analysis with a small effect size is not terribly meaningful in my opinion. I might even say worthless. Even with a larger effect size, usually the best one can say is "more research needed" unless the dataset is ruthlessly culled and any study that smells biased or not properly controlled is summarily eliminated, rather than statistically weighted.
tack is offline  
Old 07-23-2016, 11:17 AM   #16
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Quote:
Originally Posted by tack View Post
(The meta-analysis adjusted the weighting of studies that didn't control for IMD, rather than completely eliminating them. I find this spurious because that alone could account for the small effect size.)
I think this kind of statistical criticism will be interesting to follow as this study is digested. The author certainly did a lot of work to rule out various biases and confounding factors (e.g. a section that addresses publication bias (the dynamic where null/boring results are less likely to be published) and worked some statistical wizardry to see if there was still a signal despite that factor, and there was).

Quote:
Originally Posted by toddhisattva View Post
Too lazy to conduct their own experiments.
While I am skeptical of the results of this paper, I don't think you could read it and use the word "lazy" -- they/he did a lot of impressive legwork to bring the data together. Or so it appears to a layman. And as often mentioned in discussions of meta-analyses, they can have real value: the data is old, but the results can be new. And even if the results weren't new or interesting, it's still tremendously useful to the likes of us just to have someone do all the tedious research into the field and summarize all the papers and data in one place.

I really want to read a critique of it by someone versed in both audio and statistics. Hopefully that person exists somewhere and is preparing an analysis. :-)

Again, my main concern is the potentially-large proportion of incorporated studies that are paid for by financially-interested parties. Science doesn't pay for itself, I understand that. But it would have been nice to see a breakdown that considered that bias, and it's odd to me that it's not there. But I don't know enough of that world to know if that's an omission. Maybe it's not even feasible... i doubt there's much, if any, public funding to study the subject. So maybe the studies are always going to be financed by high-def audio companies.
clepsydrae is offline  
Old 07-23-2016, 11:45 AM   #17
EpicSounds
Human being with feelings
 
EpicSounds's Avatar
 
Join Date: Jul 2009
Posts: 7,568
Default

Quote:
Originally Posted by tack View Post
BTW Ian Shepherd talks about this paper in an episode of The Mastering Show.

http://themasteringshow.com/episode-17/
Thanks for sharing the podcast. Show 18 is relevant as well.

Quote:
Oh, a meta-analysis. That's all I need to know. Too lazy to conduct their own experiments.
Bull. The fact that it's a meta-analysis makes it more valuable and credible than another experiment.
__________________
REAPER Video Tutorials, Tips & Tricks and more at The REAPER Blog
EpicSounds is offline  
Old 08-04-2016, 02:28 AM   #18
zeekat
Human being with feelings
 
zeekat's Avatar
 
Join Date: May 2009
Location: Polandia
Posts: 3,578
Default

Quote:
Results showed a small but statistically significant ability of test subjects to discriminate high resolution content, and this effect increased dramatically when test subjects received extensive training.
Close listening my jams for hours to check for clipping, pops and other problems should teach me hear stuff, but still I can't even tell those allegedly apparent 128kbit mp3 compression artifacts to save my life. Hires audio seems waaaaaay outside my perception then.

(My mixes must be atrocious and I don't even know it)
__________________
AM bient, rund funk and heavy meteo
my bandcamp+youtubings
zeekat is offline  
 

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 05:42 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.