[avrosse]

Hey folks,
My father and I recently did a blind test to determine whether we could identify Red Book standard vs. high resolution music all the way up to 352.8kHz/24bit. If you’re not interested in reading about the details of the test, check out the spoiler below.

Spoiler: Warning: Spoiler!

Taking the properly volume matched test tracks as a whole, neither my father nor I performed any better than random guessing when it came to identifying whether a given track was Red Book standard or high resolution at either 192/24 or 352.8/24.
 
I’d like to stress that all this test shows is that using my specific equipment, neither my father nor I could properly determine whether a track was Red Book standard or higher resolution music. These findings do not generalize to other systems or other listeners.
 
I was quite surprised by these results and would like to encourage you all to do your own blind tests.

 
The four test tracks were:
Vivaldi: La Stravaganza, Concerto in B Flat, Op. 4, No. 1 – Allegro; Channel Classics; 92kHz/24bit
http://www.linnrecords.com/recording-antonio-vivaldi--la-stravaganza.aspx
Bach: John Passion, BWV 245, Aria mit chor - Eilt, ihr angefochtnen Seelen; Linn Records; 192kHz/24bit
http://www.linnrecords.com/recording-john-passion.aspx
Haydn: String Quartet in D, Op. 76, No. 5 - Finale – Presto; 2L; 352.8kHz/24bit
http://www.2l.no/hires/index.html and https://shop.klicktrack.com/2l/35004
Britten: Simple Symphony, Op 4: I. Boisterous Bourree; 2L; 352.8kHz/24bit
http://www.2l.no/hires/index.html and https://shop.klicktrack.com/2l/34188
 
For each of the four test tracks, a test suite of 20 tracks was produced consisting of either Red Book or high resolution versions of the test tracks. None of the tracks were either down- or up-converted by me; all tracks were purchased/downloaded from the links found above. The sequence of tracks was determined by a random number generator, the only parameter being that at least 8 of 20 tracks were either low or high rez. A sequence would look something like this: Red Book, Red Book, 192/24, Red Book, 192/24, 192/24, 192/24, Red Book etc. Each of the four test suites consisting of 20 tracks had a different sequence.
 
The test was performed using the loudspeaker setup in my house: USB Stick -> Naim DAC -> Densen B-110 -> Verity Rienzi (In case you’re wondering, a USB stick directly feeding the Naim DAC is one of the best transport mechanisms for this DAC; given the hassle associated with using USB sticks, I normally use my Oppo BDP-95 as a transport device). I chose to do the test with loudspeakers so that my father and I could do the test simultaneously. More importantly, I’ve always found it much easier to detect changes in my system using my loudspeakers than using my headphones, at least in part due to my speaker system being better than my headphone rig (Woo Audio WA6SE -> Sennheiser HD 800). Had I been able to successfully identify high rez tracks using my loudspeakers, I would’ve repeated the test using headphones.
 
 
I’ll get to the Vivaldi piece in a moment. Let’s first look at the other three test suites.
 
Percentage of correct identifications:
Bach @ 192/24:
My father: 35%
Me: 45%
 
Britten @ 352.8/24:
My father: 70%
Me: 45%
 
Haydn @ 352.8/24:
My father: 35%
Me: 55%
 
Total:
My father: 46.7%
Me: 48.3%
Taking the three tests tracks and both of our results together (for a total of 120 test tracks), we didn’t perform any better than random guessing.
 
Side note:
The Vivaldi piece was not properly volume matched. Although obviously unfortunate, it did have a very interesting upshot.
Percentage of correct identifications with the Vivaldi piece:
My father: 15%
Me: 5%
Both of us mistook the slightly louder piece to be the higher resolution version and had an easy time correctly picking it out. We both heard greater authority, fullness, detail, etc., in the louder version and just assumed that it must be higher resolution. Well, it actually turned out that the louder version was the Red Book standard! This really goes to show important it is for volumes to be perfectly matched. We didn’t repeat the test with better volume matching due to time constraints. Since this track had the lowest resolution of the high rez tracks (96/24), and considering the very clear results on the tracks with even higher resolution, I doubt that this would’ve changed the results. 
 
 
Conclusion:
Looking at the properly volume matched pieces as a whole, neither my father nor I could determine any better than random guessing whether a given track was Red Book standard or higher resolution at either 192/24 or 352.8/24. Coming into the test, I was rather confident that I wouldn’t have too much trouble doing this. Interestingly, even during the test I sometimes had the feeling that I was able to hear a difference between Red Book and high rez. Whatever differences I thought I heard, however, didn’t help me do any better than random guessing with identification. Given that you first have to hear a difference between Red Book and high rez before you can identify the track’s resolution, determining a track’s resolution would appear to be the harder task of the two, since it goes above and beyond just hearing differences. Thus, it’s possible that I heard actual differences between the two but wasn’t able to translate those differences into correct identifications. Far more likely, considering my failure to do any better than random guessing, is that I was just imagining many, if not all, of these differences.
 
I’d like to stress that all this test shows is that using my specific equipment, neither my father nor I could properly determine whether a track was Red Book standard or higher resolution music. These findings do not generalize to other systems or other listeners.
 
I was quite surprised by the results of the test, and, accordingly, I’d like to encourage you all to do your own blind tests to see whether you can correctly identify high resolution music. For those of you who want to save some money, the audiophile label 2L offers free high rez downloads at various resolutions (http://www.2l.no/hires/index.html). The Red Book standard versions of the tracks in question have to be purchased at https://shop.klicktrack.com/2l/
 
Since I’m unable to blindly identify whether a given piece of music is Red Book or high rez, I’ve decided that I won’t buy any high rez music. This decision is made even easier by the stupendous prices of many high rez downloads compared to their Red Book versions. For instance, taking a look at the four releases used for this test: Vivaldi on Channel Classics: 25 vs. 14 euros, Bach on Linn Records: 21 vs. 11 euros, Haydn on 2L: 37 vs 16 euros and Britten on 2L: 37 vs 16 euros. Not all high rez downloads are this expensive, but considering my ears and system, I personally can’t justify paying more for high resolution music.
 
Cheers,
Joel
 
 
 

 

[esldude]

Oh, yes we can generalize.  If you didn't hear the stunning obvious difference hirez makes, then you A) cannot hear worth a toot, B) had typical poor low resolution playback system components, C) had your ability to hear wiped out by the blind testing methodology or/and D) must have messed up somehow because listening sighted across multiple systems I can hear such obvious improvements with hirez there isn't even any need for blind testing.
 
 
Just kidding.......
 
I think what you did was excellent.  The test was even validated somewhat in that you perceived the volume difference.

 

[linglingjr]

  Oh, yes we can generalize.  If you didn't hear the stunning obvious difference hirez makes, then you A) cannot hear worth a toot, B) had typical poor low resolution playback system components, C) had your ability to hear wiped out by the blind testing methodology or/and D) must have messed up somehow because listening sighted across multiple systems I can hear such obvious improvements with hirez there isn't even any need for blind testing.
 
 
Just kidding.......
 
I think what you did was excellent.  The test was even validated somewhat in that you perceived the volume difference.

 
lol I was becoming really worried this would just turn into a flame war full of people trying to defend their sacd players until I got to the "Just kidding..." Thank god!
 
Great post OP if my father was ever into hi fi I would consider myself very lucky 

 
Very good

 

[xnor]

  I think what you did was excellent.  The test was even validated somewhat in that you perceived the volume difference.

Same here, appreciate it!
 
 
Even tiny volume differences have shown over and over again that usually the louder one will be preferred. Imagine how unreliable tests between DACs, amps must be when the reviewer didn't match levels properly. Not to mention sighted comparisons...

 

[nick_charles]

  Same here, appreciate it!
 
 
Even tiny volume differences have shown over and over again that usually the louder one will be preferred. Imagine how unreliable tests between DACs, amps must be when the reviewer didn't match levels properly. Not to mention sighted comparisons...

 
If you think that is unreliable how about the listeners who send off their devices to be modded and three weeks later insist that they can tell an immediate difference to what it was like 21 days prior 

 

[hogger129]

Ehh I think I can hear slightly better detail with 24/96 vs 16/44 with the same mastering; but I think 16/44 CD-quality is good enough for me.

 

[bigshot]

I think I hear better with a glass of wine, but even without wine it's good enough for me.

 

[MatsP]

I don't give much for blind tests. You can fool yourself into thinking this or that sample was better than the other from time to time, depending on the weather or what not. The most important thing is to have as much bits and sample rate available as possible, and ditch lower grade audio.

 

[gregorio]

  The most important thing is to have as much bits and sample rate available as possible, and ditch lower grade audio.

 
What makes you think that fewer bits and/or lower sample rates results in lower grade audio? Do you think, for example, that SACD (1bit) is massively "lower grade" than CD (16bits)?
 
G

 

[castleofargh]

I don't give much for blind sighted tests. You can fool yourself into thinking this or that sample was better than the other from time to time, depending on the weather or what not. The most important thing to reassure myself is to have as much bits and sample rate available as possible, and ditch lower grade audio.

admit that my edited version makes a lot more sense.

 

[spruce music]

  I don't give much for blind tests. You can fool yourself into thinking this or that sample was better than the other from time to time, depending on the weather or what not. The most important thing is to have as much bits and sample rate available as possible, and ditch lower grade audio.

Well you are getting your wish.  In the past 6 months 384 khz/32 bit DACs have become common and affordable.  Even 768khz/32 bit is not that hard to dig up. Not sure how much music is in that format yet.  I guess I am holding out for 1.536 mhz/48 bit gear.  I think we might be getting close to the transient response we need for music then.  But maybe upping sample rate is pursuing a ghost in the machine.  It may be misguided.  I think maybe simply more on the bit depth side is the way to go. 88.2 khz just to be sure and say 128 bit just to be safe.  That would be fewer bits, but put more of them where there is really sound.  We can have such tiny, tiny, tiny steps in the waveform with 128 bit.  I believe the LSB and next couple bits might toggle almost the way the single bit in DSD does.  Which would open up a new breakthrough in reproduced sound.  True, full multi-bit, DSD-like, analog sounding super precise fully self dithered audio with the resulting 576 db dynamic range.  Whew! BABY!!  With nominal 2 volt max output that LSB would trip on or off to the nearest. .000317 picopico-volts (thats trillion-trillionsth of a volt dawg). 
 
Ultra-super-mega-sonic sound quality. 

 

[gregorio]

  I think maybe simply more on the bit depth side is the way to go. 88.2 khz just to be sure and say 128 bit just to be safe.  That would be fewer bits, but put more of them where there is really sound.  We can have such tiny, tiny, tiny steps in the waveform with 128 bit.  I believe the LSB and next couple bits might toggle almost the way the single bit in DSD does.  Which would open up a new breakthrough in reproduced sound.  True, full multi-bit, DSD-like, analog sounding super precise fully self dithered audio with the resulting 576 db dynamic range.  Whew! BABY!!  With nominal 2 volt max output that LSB would trip on or off to the nearest. .000317 picopico-volts (thats trillion-trillionsth of a volt dawg). 
 
Ultra-super-mega-sonic sound quality. 

 
That's funny!! It's a bit dangerous though, such sarcasm might go over the head of some/many here on head-fi, MatsP possibly, and actually be taken seriously.
 
G

 

[Baxide]

Even tiny volume differences have shown over and over again that usually the louder one will be preferred. Imagine how unreliable tests between DACs, amps must be when the reviewer didn't match levels properly. Not to mention sighted comparisons...

There is no guaranteed way of matching volume levels. Even if you use an automatic volume level control it still won't work. All that one can do is to use a random point in the music to try to make a match. Because of the difference in dynamic range across the audio spectrum, some parts in the 24 bit audio data will be higher in value than the 16 bit red book data.
And you don't even need level matching in order to detect a difference. Every time I read it mentioned about level matching I utter a wry smile. If you ever went to buy a musical instrument and checked it out to see if it sounded better than another one, level matching would not be an issue anyone worries about.

 

[gregorio]

  There is no guaranteed way of matching volume levels.

 
Yes there is, just use an SPL meter and the same test signal going through both dacs, amps or whatever.
 

  some parts in the 24 bit audio data will be higher in value than the 16 bit red book data.

 
No they won't, the highest value for 24bit, 16bit or any other bit depth is identical, 0dBFS.
 

  Every time I read it mentioned about level matching I utter a wry smile. If you ever went to buy a musical instrument and checked it out to see if it sounded better than another one, level matching would not be an issue anyone worries about.

 
You're joking right, have you ever bought a professional musical instrument? Very subtle tonal differences and output levels make a massive difference to the cost of professional acoustic instruments. Your "wry smile" should be an embarrassed, sheepish smile!
 
G

 

[castleofargh]

 

Even tiny volume differences have shown over and over again that usually the louder one will be preferred. Imagine how unreliable tests between DACs, amps must be when the reviewer didn't match levels properly. Not to mention sighted comparisons...

There is no guaranteed way of matching volume levels. Even if you use an automatic volume level control it still won't work. All that one can do is to use a random point in the music to try to make a match. Because of the difference in dynamic range across the audio spectrum, some parts in the 24 bit audio data will be higher in value than the 16 bit red book data.
And you don't even need level matching in order to detect a difference. Every time I read it mentioned about level matching I utter a wry smile. If you ever went to buy a musical instrument and checked it out to see if it sounded better than another one, level matching would not be an issue anyone worries about.

what greg said. a multimeter, a sonometer, or a microphone. it's something many people own already even if it's in the form of a cellphone with a crappy app. the precision might not be better than 0.1db, but it's already such an improvement compared to setting things by ear. and you just use some test tone as reference when it's about testing DACs.
 
for file format, a different thing. as we all recommend very strongly to convert a highres file to lower resolution yourself instead of trying to use what's available online in different formats(because it could be another master), there should not be any loudness difference if done correctly. like not adding replay gain to the conversion or stuff like that. as I've done a few times like a noob, using preset conversion routines^_^.
and the loudest part of signal is always, as mentioned, 0db. what you increase with 24bit is dynamic range below the 16bit of CD, not withing, not above. it lets us record silence in a much higher resolution. and 24bit format is increased dynamic range for the container of the music. the music itself rarely uses more than 70db of dynamic, be it on CD or highres.