The comedy of some Hi res recordings

[andreasmaaan]

I work in the world of official statistics, so I do know the difference between anecdote and proved hypotheses. My son’s ears are anecdotes - albeit ones that he repeats all the time - but the AES study was statistically significant.

Yeh of course, I don't dispute that a few of these studies made statistically significant findings. What I'm going to need some time to sift through are the specific studies the meta-study cites that did make statistically significant findings, and what the conditions of those studies were.

For example, I've read a small part of the meta-study so far and have found that it takes into consideration:

1. studies involving bone conduction, which are not applicable to loudspeaker reproduction
2. a study that compared high-res files with standard-res files that had been rectangularly dithered, but did not include standard res files that had been dithered in the normal/correct way
3. studies whose results were later disproved
4. a widely-discussed study by Oohashi involving monitoring of subjects brainwaves, which was unable to be replicated

So I just want to see how the meta-study treated results like this, and whether its findings stand once problematic results are excluded.

And there are also my own previously mentioned concerns regarding artifacts that can be introduced downstream as a result of supersonic content. So I'd like to be satisfied that the studies that made statistically significant findings took care to avoid issues like this.

@Bluemootwo do you have an explanation for why in your opinion you and your son can distinguish standard-res from high-res and why you both prefer the latter?

 

[Bluemootwo]

Yeh of course, I don't dispute that a few of these studies made statistically significant findings. What I'm going to need some time to sift through are the specific studies the meta-study cites that did make statistically significant findings, and what the conditions of those studies were.

For example, I've read a small part of the meta-study so far and have found that it takes into consideration:

1. studies involving bone conduction, which are not applicable to loudspeaker reproduction
2. a study that compared high-res files with standard-res files that had been rectangularly dithered, but did not include standard res files that had been dithered in the normal/correct way
3. studies whose results were later disproved
4. a widely-discussed study by Oohashi involving monitoring of subjects brainwaves, which was unable to be replicated

So I just want to see how the meta-study treated results like this, and whether its findings stand once problematic results are excluded.

And there are also my own previously mentioned concerns regarding artifacts that can be introduced downstream as a result of supersonic content. So I'd like to be satisfied that the studies that made statistically significant findings took care to avoid issues like this.

@Bluemootwo do you have an explanation for why in your opinion you and your son can distinguish standard-res from high-res and why you both prefer the latter?

it’s not just me. Others on this thread have also enquired. We asked open questions about how it is 96/24 sounds better than 44/16 Some of us wondered (didn’t assert, wondered) if an analogue complex wave was sampled more frequently, perhaps the sound would be nicer to listen to.

We were told, sometimes aggressively, sometimes patronisingly, that we were the hapless victims of marketing scams and should apologise (“And it would be nice in such discussions if folks who make the initial claim you did - that a higher sampling rate brings the result "closer" to the original - would a least acknowledge that your claim, while a common-sense view, is actually mistaken.”).
So having read more than I ever have before - and I really am grateful when it was done objectively, patiently, thoroughly, and respectfully - I get the physics now I think.

Nobody has yet explained to me why my 50 year old ears, my son’s expert ears, using my very carefully built system, enjoys well-produced music more at higher resolutions than at standard ones.

if those folks here who can’t accept what many listeners do report, then having had your say your contribution is done - thanks. But I am much more interested in what others have to suggest as to why some listeners do get more enjoyment from higher resolutions. Until then, I still only have “the common sense view” available to me.

 

[Blumlein 88]

it’s not just me. Others on this thread have also enquired. We asked open questions about how it is 96/24 sounds better than 44/16 Some of us wondered (didn’t assert, wondered) if an analogue complex wave was sampled more frequently, perhaps the sound would be nicer to listen to.

We were told, sometimes aggressively, sometimes patronisingly, that we were the hapless victims of marketing scams and should apologise (“And it would be nice in such discussions if folks who make the initial claim you did - that a higher sampling rate brings the result "closer" to the original - would a least acknowledge that your claim, while a common-sense view, is actually mistaken.”).
So having read more than I ever have before - and I really am grateful when it was done objectively, patiently, thoroughly, and respectfully - I get the physics now I think.

Nobody has yet explained to me why my 50 year old ears, my son’s expert ears, using my very carefully built system, enjoys well-produced music more at higher resolutions than at standard ones.

if those folks here who can’t accept what many listeners do report, then having had your say your contribution is done - thanks. But I am much more interested in what others have to suggest as to why some listeners do get more enjoyment from higher resolutions. Until then, I still only have “the common sense view” available to me.

Were the higher resolutions you heard mastered differently? Even lower resolutions can sound anything from dreadful to sublime depending upon how they were mixed and mastered.

 

[AnalogSteph]

Yeah. I would not trust any such tests unless I had done the resampling and level-matching myself.

 

[Bluemootwo]

Were the higher resolutions you heard mastered differently? Even lower resolutions can sound anything from dreadful to sublime depending upon how they were mixed and mastered.

That could be an explanation in some cases. I know that the universal disappointment of CD listening pleasure compared to vinyl listening pleasure has driven large numbers of high-spending listeners to find alternatives to CD. Some think it is the physical media and spent more on transports. Some think it is the DAC and spent more on external DACs. Some think it is the interconnects and spent more on those too. All to no avail - whilst some improvements are reported, CD listening remains less pleasurable than vinyl to very many ears. So, maybe it was the 44/16 resolution that was at fault all along. Some music providers are going back to the masters and are creating new files to replace their 44/16 versions and sometimes when they do they get someone like Steven Wilson to produce a new master - done in and released in 96/24. (And anyone who listens to a Steven Wilson remaster is in for a real treat!) I know many audiophiles still prefer vinyl to hi res. But to my ears the hi res Qobuz streams are, generally, more pleasant to listen to than 44/16. Especially over a long listen. Many report fatigue after listening to 44/16 after an hour or so; and I do find myself engaged in pleasurable listening for much longer when using hi res. Much like I find reading much more pleasurable when a book has a better quality print - sure, I can read all the letters in newspaper print, but higher resolution folio edition prints are simply much, much nicer to read. They just are!

 

[RayDunzl]

Much like I find reading much more pleasurable when a book has a better quality print - sure, I can read all the letters in newspaper print, but higher resolution folio edition prints are simply much, much nicer to read. They just are!

If the print characters were 100 feet tall, the individual 16bit "pixels" might be about 35 thousandths of an inch square.

 

[Bluemootwo]

If the print characters were 100 feet tall, the individual 16bit "pixels" might be about 35 thousandths of an inch square.

Indeed! So help us understand how this makes listening more pleasurable, whether it is vinyl or hi res or real life or whenever the “sampling” is above 44/16.

 

[RayDunzl]

Indeed! So help us understand how this makes listening more pleasurable, whether it is vinyl or hi res or real life or whenever the “sampling” is above 44/16.

Let's make it your job to convince us, instead.

 

[Chromatischism]

Ah, car analogies.. One has to love them as they are often good, but rarely perfect!

I have a car that is capable of easilly exceeding 150mph. Maybe once or twice a year, when I'm alone in the car and the autobahn is empty I allow myself a pleasure to press the pedal to the metal and I really enjoy seeing the tachometar needle rushing toward 175mph while everything else apart from center of the road gets blurred because of speed. And then I let it go, of course, because no sane man drives like that when not on a racing course. But for that short moment, yeah.. it WAS fun.

I just zoomed into Zagreb to see what roads you could possibly use for that and ended up in a very strange "museum of broken relationships" instead. That's enough internet for tonight.

 

[Bluemootwo]

Let's make it your job to convince us, instead.

No. I and 60% of those asked (AES study) hear differences. Trained ears are more likely than that to hear differences. We just do. Asserting we can’t won’t work. If you want more people listening to more music with more pleasure (like back in the days before 44/16) then the whole industry needs to understand this. Fortunately a lot of the industry is indeed doing this - Amazon, Tidal, Qobuz - are working hard on this now. When Apple and Spotify decide they want to move on from crappy ear buds (Apple) and multi-room convenience (Spotify) then it is game over for CD and 44/16. Thirty years lost, but never mind...

 

[RayDunzl]

No. I and 60% of those asked (AES study) hear differences. Trained ears are more likely than that to hear differences.

This one?

https://www.aes.org/blog/2016/7/highresaudio

 

[Bluemootwo]

This one?

https://www.aes.org/blog/2016/7/highresaudio

Yes, that’s the one. Any scientist should be really excited by this (and not threatened by it) because examining that outcome is complex. Whether to debunk it or to explain it - complex. Much more fun than saying “you just can’t” !

 

[digitalfrost]

This here is Muddy Waters from HDTracks. Notice all the noise above 30khz...

 

[Sukie]

No. I and 60% of those asked (AES study) hear differences. Trained ears are more likely than that to hear differences.

60% of trained listeners. Not 60% of "those asked". @andreasmaaan has already pointed out a number of contributory factors to this.

There are numerous reasons why you and your son might perceive differences and these have been discussed numerous times on this forum (different mastering, different volume levels, listener bias etc.). That's why the AES Study is (potentially) important. However we can't just take their findings for granted, particularly as the study is a collation of various tests.

Fortunately a lot of the industry is indeed doing this - Amazon, Tidal, Qobuz - are working hard on this now. When Apple and Spotify decide they want to move on from crappy ear buds (Apple) and multi-room convenience (Spotify) then it is game over for CD and 44/16. Thirty years lost, but never mind...

Spotify has market-tested CD quality streaming but has, for the time being, decided against it (never mind hi-res). Tidal and Qobuz (and even Amazon HD) have a very small market share. Audiophiles might get excited about hi-res, but there is no evidence to suggest that this is shared by the general streaming public.

 

[RayDunzl]

Yes, that’s the one.

As I read the synopsis, it's a study of other's studies, a meta-study.

And he declares some studies as "flawed" if they didn't report a difference.

I wonder if those results were discarded.

I wonder if any of the studies that found a difference were flawed, and whether those results were discarded.

I wonder how his online automatic mastering service works.

"Based on the breakthroughs of his research team in intelligent music production systems, Reiss co-founded the company MixGenius in 2012. MixGenius was later renamed LandR and Reiss served as a Board member and Strategic Advisor from 2012 through 2014. LandR has received over $10 million in four rounds of venture capital funding,[9] and its online automatic mastering service has mastered over 1.4 million music tracks." --- https://en.wikipedia.org/wiki/Joshua_Reiss

 

[RayDunzl]

This one?

https://www.aes.org/blog/2016/7/highresaudio

Oh, the entire paper is available:

http://www.aes.org/e-lib/browse.cfm?elib=18296 .

(if that doesn't work go to the bottom of the page of the synopsis)

 

[Frank Dernie]

No. I and 60% of those asked (AES study) hear differences. Trained ears are more likely than that to hear differences. We just do. Asserting we can’t won’t work. If you want more people listening to more music with more pleasure (like back in the days before 44/16) then the whole industry needs to understand this. Fortunately a lot of the industry is indeed doing this - Amazon, Tidal, Qobuz - are working hard on this now. When Apple and Spotify decide they want to move on from crappy ear buds (Apple) and multi-room convenience (Spotify) then it is game over for CD and 44/16. Thirty years lost, but never mind...

I can see a few reasons why a 96/24 track and a 44/16 may sound different.

The obvious one is issued from different masters. Some of the super sounding early CD releases have been "resampled" and re-issued, but they sound very obviously different IME, with the re-sampled ones usually being complete cr*p, so ones where the difference is small would be audible.

Secondly may be if the DAC being used handles the files differently, either a different reconstruction filter or some internal resampling algorithm. That would be entirely DAC dependant.

I would not be confident any supplier advertising high res content at premium prices would be supplying CD res content from the same master - their business model relies on customers preferring their more expensive service.

To avoid either of these possibilities clouding the issue when I made a comparison I took a super quality (IMHO) 96/24 music file of music I like, Whitacre: The River Cam from his Water Night album, downsampled it to 44/16 then re-sampled it back to 96/24.
This eliminates both possibilities, firstly, it is definitely the same master and secondly, being the same file type my DAC would handle it the same way regardless of its internal workings. Of course, any data that was on the 96/24 file above 22kHz and below 16-bit depth will have been lost from the second file.
It also gives the high ground to the 96/24 file since it hasn't been manipulated at all.
Anyway, having done this I made some comparisons and the files are indistinguishable (by me). I even burnt a CD from the album, happy in the fact that it wouldn't be inferior to the 96/24 file on the computer.

I have now been listening to music for over 50 years, buying and LP a week as soon as I was earning, and also working in the record player industry, then buying CDs and being an early adopter of file based music as MP3, then MP4, then lossless at various sample rates.

I am now back on mainly CDs, and as long as it isn't a rubbish release (and sadly most are nowadays the older ones were mainly better) and I am content to be sure that there is zero audible musical information which can not be completely captured in a 44/16 file.

What DOES make a vast difference is recording quality.
I have fabulous sounding LPs and CDs but equally some awful ones.
If a recording is fatiguing it is far more likley to be the recording quality at fault than the medium on which it is distributed.

I still have a Qobuz account but, frankly, life is too short for any more comparisons in the folorn hope I may hear a minute difference between file types when there are such huge and obvious differences between recordings anyway.

One has to ask the question, if it takes a huge amount of effort to hear whether there may just be a minute difference then why not forgedabout it and get on with enjoying music?
Unless one is more interested in the technology than listening to music, that is (and I know a few people who are) .

 

[StefaanE]

I can see a few reasons why a 96/24 track and a 44/16 may sound different.
...
What DOES make a vast difference is recording quality.
I have fabulous sounding LPs and CDs but equally some awful ones.
If a recording is fatiguing it is far more likley to be the recording quality at fault than the medium on which it is distributed.
...
One has to ask the question, if it takes a huge amount of effort to hear whether there may just be a minute difference then why not forgedabout it and get on with enjoying music?
Unless one is more interested in the technology than listening to music, that is (and I know a few people who are) .

Can I like your post twice or thrice?

 

[Blumlein 88]

As I read the synopsis, it's a study of other's studies, a meta-study.

And he declares some studies as "flawed" if they didn't report a difference.

I wonder if those results were discarded.

I wonder if any of the studies that found a difference were flawed, and whether those results were discarded.

I wonder how his online automatic mastering service works.

"Based on the breakthroughs of his research team in intelligent music production systems, Reiss co-founded the company MixGenius in 2012. MixGenius was later renamed LandR and Reiss served as a Board member and Strategic Advisor from 2012 through 2014. LandR has received over $10 million in four rounds of venture capital funding,[9] and its online automatic mastering service has mastered over 1.4 million music tracks." --- https://en.wikipedia.org/wiki/Joshua_Reiss

I put in a few tracks worth during a free trial period at LandR. It wasn't great, not completely terrible. I can't see it doing very well unless it improves. Then again it has lasted longer than I expected.

 

[Blumlein 88]

There is this interesting test done a few years ago regarding 44.1 vs 88.2 khz rates. PDF attached to the other post.
https://www.audiosciencereview.com/...e-rate-and-audible-frequency.9411/post-247125

Very good gear, capable of extended response. Blind test. In short 88.2 khz resampled to 44.1 khz was audibly different. But 88.2 khz vs 44.1 khz recording done concurrently was not audibly different at less than 5% chance levels. Apparently resampling was audible. It was Izotope used for that and it is one of the better ones.

Don't know if you've seen Arny's jangling keys recording that was around for years. I could abx that at the different rates pretty easily. He took a high rate recording and downsampled. When I took his original hi-rate file and resampled with Sox, I could no longer hear a difference anymore. His resampler whatever it was apparently wasn't great. Quite a number of resamplers built into DAWs still aren't very good.

If a good recording hasn't been processed to death, I prefer to hear it in its native sampling rate.

There are few reasonable ideas as to why one might wish to record at 88 or 96 khz rates. There might, maybe, could be, in some circumstances an audible benefit. Even the meta-analysis mentioned above indicates only a mild audibility if any.

The difference in anything beyond 48 khz is very small. If it weren't we'd have long ago determined it is for certain better. Considering all the various factors it isn't surprising you might have a fat gray line instead of a single sharp one as to where more is not better. 44/16 probably captures at least 98% of what is possible. If sourced from 24 bit files and done impeccably well on both ends of the conversion it might be 100% of all that is necessary. So 88 or 96 khz recordings don't bother me too much, and I'd like that rate if it was the native rate of recording. More than 96 khz however is definitely just a load of malarkey. The idea well done 44/16 by itself is an annoying medium is equally malarkey. Won't you don't hear doesn't create problems for your hearing that makes it an unenjoyable medium. Poor recording and mastering does that, and does it with any medium.