Is high-resolution audio audible or not audible and a waste of data?

[Talisman]

Even when I listen to music in the car or with the iPod, I can hardly tell if it's mp3 quality (naturally with little compression) or CD, and I'm happy about it. But these are not listens that I consider "quality". Comparisons must be made on quality playback chain, controlled environments and above all good audio material, with ABX tests and more persons.

Of course I was referring to careful listening, either with my main system or with my headphones.
If we also include a car system heard on the street, well then I'd be surprised if someone recognized even a 128kbs mp3 from a flac.
I personally did the proposed ABX tests which play lossless files and then mp3 compression at various levels. I can tell you that already at 128kbs you have to be seriously careful, and for me at 320kbs the files were indistinguishable.
Many people's requirement doesn't interest me, for me it's not a scientific study on what the human ear can perceive, but only on my personal limits.
Knowing my limits helps me make more informed choices and better direct my economic resources

 

[voodooless]

The fact that our auditory system could detect this effect in the form of an alteration of the localization of the sources in the space.

Citation needed! Is this a fact? In what way? What research concluded this? You say “could”, not “can”. “Could” implies probably, not certainty.

I surely hope your not going to quote any Milind Kunchur…

 

[antcollinet]

This is something that puzzled me too at first. The 20KHz bandwidth is related to the spectral analysis of the sound performed by the cochlea, to determine its "timbre". But for localization aspects, for which the differential analysis of arrival time and/or signal levels (ITD and ILD) counts, this limit does not apply. In modern psychoacoustic texts these mechanisms are well explained.

As others have pointed out - tiny arrival time differences of already band limited signals can still be detected.

 

[Pinox67]

Citation needed! Is this a fact? In what way? What research concluded this? You say “could”, not “can”. “Could” implies probably, not certainty.

I surely hope your not going to quote any Milind Kunchur…

If you read the texts that talk about these experiments, you can see that the ITD/ILD values that we are able to detect depends on several aspects, such as the frequencies in the sound, spectral complexity, background noise. And it doesn't seem like a surprise to me.

 

[Pinox67]

As others have pointed out - tiny arrival time differences of already band limited signals can still be detected.

Correct, the temporal aspects of the transients are preserved in this case. I ran some simulations too.

 

[voodooless]

If you read the texts that talk about these experiments, you can see that the ITD/ILD values that we are able to detect depends on several aspects, such as the frequencies in the sound, spectral complexity, background noise. And it doesn't seem like a surprise to me.

What text? This is not a citation... And what does this have to do with the sample rate? The transients you describe are invalid signals. They are irrelevant in real music. Digitally sampled signals do not have discrete phase.

 

[Pinox67]

What text? This is not a citation... And what does this have to do with the sample rate? The transients you describe are invalid signals. They are irrelevant in real music. Digitally sampled signals do not have discrete phase.

I can't find the main articles on the fly. However you can search the web for "Interaural Time Difference" (or Level). As already reported, you can also find a lot in good psychoacoustic books. A very interesting one that I suggest is "Principle of Cognitive Neuroscience", by D. Purves and others.

[EDIT] About the test signal. As normally happens, to study the effects of any operation on the signal, tests are carried out with "canonical" signals. The step is one of them. Yes, it's not a realistic signal, but it shows well the time spreading effect of the transients, which is a physical fact. I think that on real signals surely such steep transients will occur to a lesser extent.

For discrete phases. It doesn't seem to me that anyone here has talked about discrete phases.

 

[voodooless]

I can't find the main articles on the fly. However you can search the web for "Interaural Time Difference" (or Level). As already reported, you can also find a lot in good psychoacoustic books. A very interesting one that I suggest is "Principle of Cognitive Neuroscience", by D. Purves and others.

So where is the research that links sample rate to "soundstage" and "fatigue"?

 

[danadam]

Download samples from 2L as they have multiple resolutions of entire demo tracks.

Sadly not available anymore. Maybe on https://web.archive.org

 

[Blumlein 88]

I can't find the main articles on the fly. However you can search the web for "Interaural Time Difference" (or Level). As already reported, you can also find a lot in good psychoacoustic books. A very interesting one that I suggest is "Principle of Cognitive Neuroscience", by D. Purves and others.

So what exactly are you surmising from this? What level of interaural time difference is required? It is 10 microseconds or slightly less and redbook CD can do 110 picoseconds or roughly an interval 90,000 times shorter.

 

[danadam]

I don't know if anyone is capable of feeling such differences.

Here are sample files with differences increasing by 5 µs: https://hydrogenaud.io/index.php?topic=107570.msg899713#msg899713
Of course there's no hi-res involved here. In fact the content is low-passed to 2 kHz.

Here's my 15/15 ABX log for 20 µs:

Spoiler

foo_abx 2.0.2 report
foobar2000 v1.3.10
2017-11-22 02:35:21

File A: Impulses shift 0 samples 2klp norm 4416 .flac
SHA1: 8fc00a4bb6a1bb0a66ec5c83cfaa36f9d8fddd13
File B: Impulses shift 4 samples 2klp norm 4416 .flac
SHA1: 6133aaa124c97a3f768f3d9216af2eb07b7c0bf3

Output:
ASIO : ASIO4ALL v2
Crossfading: NO

02:35:21 : Test started.
02:38:48 : 01/01
02:39:48 : 02/02
02:40:31 : 03/03
02:42:31 : 04/04
02:44:00 : 05/05
02:44:54 : 06/06
02:47:26 : 07/07
02:49:35 : 08/08
02:56:36 : 09/09
02:57:49 : 10/10
02:59:46 : 11/11
03:01:23 : 12/12
03:03:16 : 13/13
03:04:04 : 14/14
03:07:30 : 15/15
03:07:30 : Test finished.

----------
Total: 15/15
Probability that you were guessing: 0.0%

-- signature --
b76d1c64aff74736f104c5cd98c431b6c12b6dc8

And here someone had 8/10 for 10 µs: https://hydrogenaud.io/index.php?topic=107570.msg899732#msg899732

 

[Waxx]

Even on a high quality system, 44.1kHz and 16bit is beyond what you can hear. The speakers, even the best ones in the world have a lower resolution. 16bit is a dynamic range of 96dB theoretical, and no speaker has that dynamic range. And nobody hears above 20kHz, even not the best ears. Most of adult age even don't hear above 16kHz. So that kind if High res is mostly a waste of HD space i think (altough i also own some music in 192kHz 24bit or 96kHz 24bit format). Mp3 or other lossy formats don't do it, but when it's lossy and that format or higher, there is no difference I and most think based on many studies done about this subject.

 

[ZolaIII]

Cuple years ago I participated in a survey with hires 24 bit and CD quality track's with A-B testing. I could identify the difference in 75% of track's listening on above average (90+ dB program) listening level's fully concentrated using a good hedaphones but it whose barely auditable even listening like that. Being puzzled and offended with golden eyes comments I did series of transcoding tests using DB power converter to minimise impact of sample rate quantitation errors and doing it all on single CPU core (no slicing) It turned out that all of lossy compression formats failed (MP3, AAC, Opus) as they had their own limits regarding down sampling and used their own in built algorithms to do it. I used both graphs and hearing so it whosent subject related in any other way than verification.
However WavPack lossy on relatively high bitrate shaving only cuple top bit's (4~5 from 24) and not doing resampling whose a hit with no perceivable difference in graphs or listening practice. The compression ratio compared to flac for such high sample rate whose 5:1 or more.
True is that most (more than 90%) of so called hires tracks are nothing other than upsampled CD or lower quality one's and that it all together isn't worth a fuss.
I don't even want to mention MQA which is just a propetry ripoff. In other words CD quality is fine and more than 90% music is in it and for very long it will stay like that. We audibly can go up to 20 KHz and 19~20 bit's while best DAC's stretch 22 bit's. And even so most of us won't be able to hear more than wide band (16 KHz) and there is little above normal broad band (12.5 KHz) in recorded music materials but we are sensitive to resampling errors. This actually means that there is lot that can be done regarding making better lossy compression formats and if done properly you can save significant amount of bandwidth preserving fidelity we actually can hire. Take WavPack lossy with metadata as one of existing one's for example for actual streaming (lossy) and storage with metadata so that it can be reconstructed (metadata actually contains all which is shaved off) back to original losseles source master. The compression ratio for 44100 Hz 16 bit materials won't be a big gain compared to Flac ( about 1:1.35~1.65) but if you are a on demand streaming service that's actually a lot.

 

[Blumlein 88]

Even on a high quality system, 44.1kHz and 16bit is beyond what you can hear. The speakers, even the best ones in the world have a lower resolution. 16bit is a dynamic range of 96dB theoretical, and no speaker has that dynamic range. And nobody hears above 20kHz, even not the best ears. Most of adult age even don't hear above 16kHz. So that kind if High res is mostly a waste of HD space i think (altough i also own some music in 192kHz 24bit or 96kHz 24bit format). Mp3 or other lossy formats don't do it, but when it's lossy and that format or higher, there is no difference I and most think based on many studies done about this subject.

I know it is nit picking, but actually there are people who hear to around 25 khz. 1% or 2% of young adults (say below 30 years old) can hear a bit over 20 khz with a few of them hearing to 25 khz. It is confirmed in actual tests. They do have exceptionally high thresholds for this perception like 100 db or higher.

 

[xaxxon]

I can tell you that already at 128kbs you have to be seriously careful,

Depends on the contents of the song. hihats are trivial at 128

Do 320kbps mp3 files really sound better? Take the test! - Audio and Sound

A while ago, I decided to switch to MP3 music instead of CD's, so I painstakingly ripped all my CD's (500+) onto my computer. It's much easier finding albums on a computer than it is sifting through piles of CD's only to find out that I put the wrong CD in the case that I

www.noiseaddicts.com

I just got this right. All I had to do was listen to the hats.

The site seems to have an error in their code, but the answer is still on the page.

 

[Waxx]

I know it is nit picking, but actually there are people who hear to around 25 khz. 1% or 2% of young adults (say below 30 years old) can hear a bit over 20 khz with a few of them hearing to 25 khz. It is confirmed in actual tests. They do have exceptionally high thresholds for this perception like 100 db or higher.

I was tested at the University of Ghent (Belgium) because i could hear "exceptional" high they said, but my first test at the age of 10 was just above 19kHz. They started to test me because i complaint about high pitched sounds of old tv's and other electronic devices that nobody hears. Never heared about people who hear above 20kHz. Do you have links to those studies?

Btw, now at the age of 43 i still can hear until 16854Hz (test done a few months ago). And that is also exceptional they say at the university, especially for a former dj and sound engineer who worked often in very loud envirroments. But i don't see myself as a golden ear, it's more a problem because i hear all kind of stuff i don't want to hear. I can also hear very silent sounds, that normal humans can't, and that is a big problem. I don't want to hear the insects that live in my house (in every house) when i'm in my bed, i want silence.

 

[voodooless]

Depends on the contents of the song. hihats are trivial at 128

Do 320kbps mp3 files really sound better? Take the test! - Audio and Sound

A while ago, I decided to switch to MP3 music instead of CD's, so I painstakingly ripped all my CD's (500+) onto my computer. It's much easier finding albums on a computer than it is sifting through piles of CD's only to find out that I put the wrong CD in the case that I

www.noiseaddicts.com

I just got this right. All I had to do was listen to the hats.

The site seems to have an error in their code, but the answer is still on the page.

Hmm, "Made with Sony ACID Pro 6.0". Not sure how good that encoder is.

 

[Pinox67]

So what exactly are you surmising from this? What level of interaural time difference is required? It is 10 microseconds or slightly less and redbook CD can do 110 picoseconds or roughly an interval 90,000 times shorter.

I think there is some confusion.
One thing is that for a signal limited in audio bandwidth, the sampling rate of 44KHz is able to reproduce with high precision the instant in which a transient occurs (110psec, although I should elaborate on how this value is calculated), which exceeds the level of temporal accuracy (ITD/ILD) of our ear.

Another thing is that the transients whose rise rate is lower than 22usec are diluted in time, before and after the event, as reported in the graph of the first post. And since our ear has a higher sensitivity than this value (6-10usec), 44KHz is not physically sufficient to preserve transient events for our ears.

The other issue is then on whether or not these speeds are present in real musical signals, the recordings are able to record them, the reproduction systems are able or not to return such precision, the environment, etc. all factors that can affect listening.

 

[Gorgonzola]

For me at least, given I'm totally deaf above about 9 kHz, high-rez serves no purpose.

Furthermore & more importantly, IMO, the quality of the recording and mastering has far more to do with sound quality of the recording than does resolution.

 

[danadam]

Another thing is that the transients whose rise rate is lower than 22us are diluted in time, before and after the event, as reported in the graph of the first post. And since our ear has a higher sensitivity than this value (6us-10us), 44KHz is not physically sufficient.

I just don't see how you can equate sensitivity to ITD with sensitivity to rise time.