Errors of Mis-Placed Precision in Audiophility

[Ralf11]

Based on some recent repartee, I thought this might be a useful thread ...

 

The opening title phrase refers to a common procedure done by students in STEM classes - they will incorrectly put too many decimal places after a number for a measurement (in the most simple example).  This denotes a more precise measurement than was actually made.

 

I've seen similar things done in related contexts - for example, in one lab great effort was made to increase the precision of oxygen measurements - when the real issue was the volume of gas flow (that and some other issues lead to an NIH site visit and eventual loss of a career by the not so technically astute scientist involved).

 

One thing for audiophiles is to consider if higher quality (and often, but not always) more expensive gear will really enhance the sound.  One situation where it may not is where the source recordings one wishes to hear are not of high SQ to begin with.  Examples include very early ragtime, jazz, blues, and Dead shows recorded on cassettes.

 

Other situations are where an improvement in - say - noise or power supplies do not affect the signal carried like, arguably, the Apple TV).  Certain cables often fall into this category as well (AC power cables, speaker cables for Magneplanar speakers, overly expensive speaker cables, interconnects, and some others).

[esldude]

32 bit DACs?  SOTA electronics can manage maybe 130 db or more often a good 120 db (20-21 bits).  If you wish to listen to music as loud as 120 db, you'll have noise 10 db below the hearing threshold in rough terms.  So instead of the 144 db that can be addressed by 24 bits we need 32 bit DACs that could address 192 db, except of course no electronics are anywhere close to that quiet.  What is the point? Oh, and should I mention much music has less dynamic range than ever in the history of recorded music.  

 

Sample rates above 96 khz?  768 khz DACs are starting to be almost common (384 khz already is not unusual).  Why the increasingly common 384 khz rates and now 768 khz rates when devices to record at those numbers are sparse?

 

Somewhat similar to megapixel wars in cameras.  I would settle for 96 khz 24 bit devices with really nice analog circuitry and no spurious tones or aliasing or such issues.  Just like a good 6 to 10 megapixel camera with a nice large low noise sensor would be very nice to have. 

 

Oh and let us not forget word clocks external to the device.  And the obsession with really, really, really, really, really precise clocking.  Plenty of devices should scrap the money on that and put $20 toward the analog circuitry involved (within which I include the power supplies).  

[Allan F]

Audiophility!?!? And I suppose your DAC plays digitality?

[Ralf11]

you're being sillity

[synn]

A rig can only be as good as the source material.

 

it doesn’t matter if one has a 768khz 32 bit DAC if all they do is play Kanye West and Taylor Swift (or the likes).

 

I have Red Book quality downloads from Linn Records and B&W Society of Sound that sound astounding on a half decent rig. Plus of course the Dire Straits CDs. High quality high res material of course has that little bit (admittedly incremental) extra. 

 

A lot of my source material is at red book or 48khz 24 bit, some at 96 and a few albums at 192. Which is why I went for a quality DAC which does 192 REALLY GOOD instead of something from SMSL etc that has way higher specs on paper (no disrespect to SMSL, I have heard that they make some fine sounding gear).

[Ralf11]

I sometimes hear that some equipment is too revealing - you hear every flaw...

 

Many listeners have a mix of source quality - e.g. recent jazz along with stuff from the beginning of recorded music...

[NOMBEDES]

Most folks think wine in a heavy bottle is better than wine in a standard bottle.  What you are getting is more glass, not better wine.

I have never been able to find any improvement in sound with higher resolution, as stated above by @synn, if the source material is good, the music sounds good.   

 

 

[fas42]

1 hour ago, NOMBEDES said:

Most folks think wine in a heavy bottle is better than wine in a standard bottle.  What you are getting is more glass, not better wine.

I have never been able to find any improvement in sound with higher resolution, as stated above by @synn, if the source material is good, the music sounds good.   

 

 

 

With a competent system, CD resolution is way beyond enough to deliver the goods - hires is just a workaround for playback that has implementation problems. A simple solution for such a setup is just to upsample RB to the "best" res for the rig beforehand, and play those instead.

[Ralf11]

the most recent study on this was a meta-analysis that found small significance in tests of hi-res vs. Redbook

 

Reiss, J. D.  2016.  A Meta-Analysis of High Resolution Audio Perceptual Evaluation.  JAES 64(6): 364-379.  

 

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

[NOMBEDES]

6 hours ago, Ralf11 said:

the most recent study on this was a meta-analysis that found small significance in tests of hi-res vs. Redbook

 

Reiss, J. D.  2016.  A Meta-Analysis of High Resolution Audio Perceptual Evaluation.  JAES 64(6): 364-379.  

 

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

 

So I can buy “ Kind of Blue” at the used CD store for $5 or I can buy a high Rez download for $25.  I am guessing the “small significance” is not worth the extra money.  If you can hear the difference, go for it.

[Ralf11]

Not if the used CD has a different recording/mastering than the Hi-Res release - and usually the latter has a lot of care taken

[firedog]

3 hours ago, NOMBEDES said:

 

So I can buy “ Kind of Blue” at the used CD store for $5 or I can buy a high Rez download for $25.  I am guessing the “small significance” is not worth the extra money.  If you can hear the difference, go for it.

Uh, the KOB hi-res remasters are also better sounding even if you convert them to Redbook. They simply are a superior mastering done with more care and attention to detail than the typical versions you come across. Anyone can hear a difference. How much money that is worth to you is a different question. 

[Miska]

On 12/5/2017 at 3:46 AM, esldude said:

32 bit DACs?  SOTA electronics can manage maybe 130 db or more often a good 120 db (20-21 bits).  If you wish to listen to music as loud as 120 db, you'll have noise 10 db below the hearing threshold in rough terms.  So instead of the 144 db that can be addressed by 24 bits we need 32 bit DACs that could address 192 db, except of course no electronics are anywhere close to that quiet.  What is the point? Oh, and should I mention much music has less dynamic range than ever in the history of recorded music.  

 

And it is good to count for some 30 dBA background noise in a listening room...

 

On 12/5/2017 at 3:46 AM, esldude said:

Sample rates above 96 khz?  768 khz DACs are starting to be almost common (384 khz already is not unusual).  Why the increasingly common 384 khz rates and now 768 khz rates when devices to record at those numbers are sparse?

 

External oversampling? Why would oversampling process need to be inside the DAC and offer only low sampling rates as input?

 

So far all DACs I've seen to do 8x oversampling inside still leak images indicating incomplete reconstruction. Making an analog filter to fully reconstruct signal in frequency domain will screw up phase response badly in audio band...

 

[Miska]

One thing I would add to the list:

 

DACs that do DSP inside instead of just simply converting input data into analog as well as possible. Computers are anyway used for playback and can do the DSP stuff better than what is inside DACs, without added cost. So the in-DAC DSP is just stealing money from the parts that really matter, which is the actual conversion section and analog stages.

 

[Miska]

11 hours ago, Ralf11 said:

the most recent study on this was a meta-analysis that found small significance in tests of hi-res vs. Redbook

 

Reiss, J. D.  2016.  A Meta-Analysis of High Resolution Audio Perceptual Evaluation.  JAES 64(6): 364-379.  

 

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

 

This is one the many bogus AES publications. Were the subject persons trained in listening for differences? And in many cases the equipment used is not suitable for hires in first place.

 

I know how much listening training can make a difference after training passive sonar operators for many years. When they are fresh they are not able to detect almost anything from background noise, and couple of months of training and experience they are soo much better listeners already...

 

My primary tool for listening differences is Sennheiser HD-800 headphones.

 

[NOMBEDES]

7 hours ago, firedog said:

Uh, the KOB hi-res remasters are also better sounding even if you convert them to Redbook. They simply are a superior mastering done with more care and attention to detail than the typical versions you come across. Anyone can hear a difference. How much money that is worth to you is a different question. 

 

You well may be correct that KOB was remastered with more care, however, there are thousands of complaints posted on various audio/music sites concerning upsampled music passed off as high quality, high resolution, higher cost product.

As I stated above, if you can hear a difference......enjoy! 

[Samuel T Cogley]

9 hours ago, firedog said:

Uh, the KOB hi-res remasters...

 

Which one?  Serious question.  Isn't this one of the most remastered titles on the planet?

[Speedskater]

19 hours ago, Ralf11 said:

the most recent study on this was a meta-analysis that found small significance in tests of hi-res vs. Redbook

Reiss, J. D.  2016.  A Meta-Analysis of High Resolution Audio Perceptual Evaluation.  JAES 64(6): 364-379. 

There were questions about the suitability and validity of many of the tests used in this meta-analysis.

[Fitzcaraldo215]

7 hours ago, Miska said:

 

This is one the many bogus AES publications. Were the subject persons trained in listening for differences? And in many cases the equipment used is not suitable for hires in first place.

 

I know how much listening training can make a difference after training passive sonar operators for many years. When they are fresh they are not able to detect almost anything from background noise, and couple of months of training and experience they are soo much better listeners already...

 

My primary tool for listening differences is Sennheiser HD-800 headphones.

 

You should read it.  I think it is quite a good summary of available research, even if one does not share its conclusions.

 

To answer your question, a section is devoted to a comparison of results between trained and untrained listeners, to the extent that different test methods were known to have done that training. The better discrimination in tests using trained subjects is fairly clear. And, I agree that training in advance for DBTs of any kind is important.  Without it, the test results might be less reliable and compelling, particularly when small differences are involved.

 

People will look at that paper and its conclusions in different ways, perhaps justifying their prior beliefs.  I, for one, find justification in it supporting my preference for hirez.  Others might not.

[firedog]

54 minutes ago, Samuel T Cogley said:

 

Which one?  Serious question.  Isn't this one of the most remastered titles on the planet?

I think we were referring to the high res ones from a few years ago. They are the most recent digital ones, other than a DSD remaster.

[Ralf11]

13 hours ago, firedog said:

Uh, the KOB hi-res remasters are also better sounding even if you convert them to Redbook. They simply are a superior mastering done with more care and attention to detail than the typical versions you come across. Anyone can hear a difference. How much money that is worth to you is a different question. 

 

download only?  or is there an SACD that is superior?

[Ralf11]

3 hours ago, Speedskater said:

There were questions about the suitability and validity of many of the tests used in this meta-analysis.

 

I'm not aware of any publications to that point - can you cite them?

 

For the meta-analysis, they removed quite a few studies.

 

I do agree more studies would be useful

[firedog]

3 hours ago, Ralf11 said:

 

download only?  or is there an SACD that is superior?

Can’t remember all the details. I’m sure you can find the information if you look. 

[Speedskater]

18 hours ago, Ralf11 said:

 

I'm not aware of any publications to that point - can you cite them?

For the meta-analysis, they removed quite a few studies.

I do agree more studies would be useful

About 400 hundred posts on the "A Meta-Analysis of High Resolution Audio Perceptual Evaluation" paper.

https://hydrogenaud.io/index.php/topic,112204.0.html

[Speedskater]

Another discussion of the "A Meta-Analysis of High Resolution Audio Perceptual Evaluation" paper. 

https://www.audiosciencereview.com/forum/index.php?threads/aes-meta-analysis-on-audibility-of-hi-res.604/