How Well Can You Hear Audio Quality?
Jun 11, 2015 at 9:58 AM Thread Starter Post #1 of 17

ambchang

500+ Head-Fier
Joined
Feb 4, 2008
Posts
697
Likes
56
http://www.npr.org/sections/therecord/2015/06/02/411473508/how-well-can-you-hear-audio-quality
 
Got this test online, and I gotta say, I am horrible at this.
 
I first used the direct computer audio out from my Dell to a BA200, I can't tell a difference one way or the other.
 
Then I used the Stoner Acoustic UD100, to Beta22 amp to HE500 (I know, I need to upgrade my source), and while I can tell the 128kps sound for 5 out of the 6 tests (ironically, I couldn't tell the difference for the Neil Young song when he came up with Pono), but I CONSISTENTLY picked 320 kps music over WAV, with the exception of the classical music.
 
I can hear that the 128k music is the worse quality, and that is somewhat apparent to me, but I can only tell there are slight differences between 320k and wav, and even then, I prefer 320k.
 
Man, I am really bad at this.
 
If interested, post and see how you did and share your experiences.
 
Jun 13, 2015 at 12:47 PM Post #2 of 17

cel4145

Headphoneus Supremus
Joined
Dec 16, 2011
Posts
19,500
Likes
2,316
Location
grand rapids, mi
Some discussion already of that test starting here if you are interested: http://www.head-fi.org/t/769647/objectivists-board-room/360#post_11675462
 
Jun 13, 2015 at 2:23 PM Post #3 of 17

blades

100+ Head-Fier
Joined
Jun 8, 2014
Posts
170
Likes
28
I've gone through formal bias controlled comparisons.  My results are that 256K and 320K MP3 sound the same to me as the native CD.  With some recordings I can distinguish between the CD and a 192K MP3.  At 128K I can distinguish them every time.
 
Jun 14, 2015 at 3:17 AM Post #4 of 17

arnyk

Repeatedly defended arguments with personal attacks.
Joined
May 30, 2015
Posts
737
Likes
116
  http://www.npr.org/sections/therecord/2015/06/02/411473508/how-well-can-you-hear-audio-quality
 
Got this test online, and I gotta say, I am horrible at this.
 
I first used the direct computer audio out from my Dell to a BA200, I can't tell a difference one way or the other.
 
Then I used the Stoner Acoustic UD100, to Beta22 amp to HE500 (I know, I need to upgrade my source), and while I can tell the 128kps sound for 5 out of the 6 tests (ironically, I couldn't tell the difference for the Neil Young song when he came up with Pono), but I CONSISTENTLY picked 320 kps music over WAV, with the exception of the classical music.
 
I can hear that the 128k music is the worse quality, and that is somewhat apparent to me, but I can only tell there are slight differences between 320k and wav, and even then, I prefer 320k.
 
Man, I am really bad at this.
 
If interested, post and see how you did and share your experiences.

 
The NPR test is IMO pretty horrible. It is prone to both false positives and false negatives.
 
The right way to do this sort of thing involves the use of a software ABX Comparator such as FOOBAR2000 with the ABX plug in (freeware).
 
That in turn requires that the test files be separately downloadable.
 
Jun 21, 2015 at 10:33 AM Post #7 of 17

inthere

Headphoneus Supremus
Joined
Apr 11, 2012
Posts
2,468
Likes
705
  Listened to them all, wrote the answers down,  then took a screen shot.
This is for people who say nobody can tell the difference.

  Some people (not me!) will tell you that all that shows is that you had the answers memorized. 
 
Jun 21, 2015 at 11:40 AM Post #9 of 17

arnyk

Repeatedly defended arguments with personal attacks.
Joined
May 30, 2015
Posts
737
Likes
116
  Some people (not me!) will tell you that all that shows is that you had the answers memorized. 

 


I don't know why.

The alternatives seem to be primarily things that are generally accepted by most knowledgeable people to sound different.

It is generally accepted that there are audible differences between .wav flies and any of the different flavors of perceptual coders (AAC, MP3, etc) for example as well as different bitrates of the same coders.

If you can memorize how some alternative file sounds, then that counts as an audible difference.
 
Jun 26, 2015 at 8:23 PM Post #10 of 17

asdfg

WARNING: BAD TRADER
Joined
Jan 2, 2007
Posts
55
Likes
13
Definitely, this is somewhat a subjective issue, very hard to debate about
tongue_smile.gif

 
Jun 27, 2015 at 12:39 AM Post #11 of 17

ProtegeManiac

Headphoneus Supremus
Joined
Oct 29, 2009
Posts
16,228
Likes
3,045
Location
Manila
  http://www.npr.org/sections/therecord/2015/06/02/411473508/how-well-can-you-hear-audio-quality
 
Got this test online, and I gotta say, I am horrible at this.
 
I first used the direct computer audio out from my Dell to a BA200, I can't tell a difference one way or the other.
 
Then I used the Stoner Acoustic UD100, to Beta22 amp to HE500 (I know, I need to upgrade my source), and while I can tell the 128kps sound for 5 out of the 6 tests (ironically, I couldn't tell the difference for the Neil Young song when he came up with Pono)...
 
...
I can hear that the 128k music is the worse quality, and that is somewhat apparent to me, but I can only tell there are slight differences between 320k and wav, and even then, I prefer 320k.
 
Man, I am really bad at this.

 
My internet is throttling right now since it's a weekend here (lots of people streaming at home, usually), but I generally have one issue against "unbiased" listening tests like these. Other tests for example use tracks from audiophile albums (and thankfully not this one), including one of the earliest that featured a xylophone solo, and all of them are almost all midrange tracks that barely have any information in the range that is affected by MP3 compression in the first place. This one uses a wider variety, but there's still the problem of whether one is familiar with the music to begin with.
 
As much as there's the risk of placebo and lack of blind testing, using one's own music has one advantage in that you know it well enough. In my case what you can get away with using 256kbps and what still has some kind of difference even on 320kbps aren't what people would expect. Audiophile albums can more easily get away with 256kbps (as above), but metal (broadly speaking, but I mostly listen to power, symphonic, and prog) won't, because it covers a wider range of music. Whereas there's that audiophile xylophone solo or the typical vocal and piano track used in earlier tests which have no percussion much less bass drums, metal has a full plethora of instruments covering a wider spectrum, including tracks that include a full brass and string section. Double-pedal bass drum action has a tendency to sound like you need to stuff it with a pillow on 256kbps, or some of the overall tone of the backing orchestra sounds different.
 
As much as this article isn't as biased as the one that used a xylophone solo, if you don't listen to these tracks regularly and not in lossless either, there's no way to tell one track from the other, and what it proves is that, in practice, listening to an unfamiliar track, you can't tell the difference. Similarly, as I said that audiophile tracks can get away with compression easier, maybe someone who regularly listens to those tracks might be able to say if this holds true for those tracks as well.
 
Basically, my point is that familiarity with the music and particular aspects of each track or genre can make the effects of compression vary from one to another. And personally, for most music out there, in practice people more likely wouldn't, so don't sweat it. There's no golden ear award if you can nor a pariah punishment if you can't, because hey, it's not necessarily your ears, but the range covered by the track, and placebo would be worse 
biggrin.gif

 
 
...but I CONSISTENTLY picked 320 kps music over WAV, with the exception of the classical music.
 
I can hear that the 128k music is the worse quality, and that is somewhat apparent to me, but I can only tell there are slight differences between 320k and wav, and even then, I prefer 320k.

 
Like I said above, it depends on the range, and some of the more epic classical compositions have a very wide range of frequencies in them. Also, it can depend on the overall system response. If you're using a system with a lot of bass boost or even also some treble boost, then that can make a properly recorded track bloated and sharp, but somewhat (note: not exact) evens out on 320kbps.
 
 
EDIT: There is one flaw in the test for those with slow internet speeds - whichever one doesn't start right away or stutters is the WAV file. I got most of them correct, or when wrong it was still the 320kbps. I cycled through them several times but I think that delay can't be ignored; there's that one track that was just one female voice and I'm damn sure that without such a cue I would not get that right. In one of the tracks that I got wrong, "Tom Ford" by Jay-Z, I got it wrong because I momentarily forgot that it was about which track was WAV as opposed to which one I liked best, and the overall bass response (thanks to bass boost on my IEM) was tighter on the 320kbps, slightly bloated on WAV, and was slightly weaker but still a bit "loose" on the lowest quality track. 
 
Jun 28, 2015 at 12:46 PM Post #12 of 17

judgmentday

100+ Head-Fier
Joined
Feb 18, 2003
Posts
450
Likes
12

This is like the effect of testing 2 different colognes at the same time. After a while both smell the same.  Some how the brain refuses to catch differences but there are big differences.
The compressed file sound thin, bright, no separation, no space, instruments sound smaller, no real bass and after a while of listening one gets a headache or ear pain.  I hate digital recordings anyways. I'm an analog purist when it comes to listen to music seriously. Still I listen to FLAC or MP3 @ 320kb on my computer when working.
 
Jun 28, 2015 at 3:07 PM Post #13 of 17

nick_charles

Headphoneus Supremus
Joined
Feb 26, 2008
Posts
3,180
Likes
334
 
This is like the effect of testing 2 different colognes at the same time. After a while both smell the same.  Some how the brain refuses to catch differences but there are big differences.
The compressed file sound thin, bright, no separation, no space, instruments sound smaller, no real bass and after a while of listening one gets a headache or ear pain.  I hate digital recordings anyways. I'm an analog purist when it comes to listen to music seriously. Still I listen to FLAC or MP3 @ 320kb on my computer when working.

 
I took a classical CD and ripped a 30 second segment as wav and as a 320k mp3. Then I scoped out the FRs as below, click to expand. The FRs were very close all the way up to 16K then the MP3 rolls off, if anything this would make the uncompressed file brighter/thinner not make the MP3 sound thinner as the bass regions are very close compared to the high end which is relatively hard to hear anyway. In the low bass region from 22 - 199 hz the average difference between the two was -0.0374 db. the differences at any frequency did not exceed 0.6db in any direction (Min -0.6, max +0.6) overall the MP3 had slightly more bass energy. Approximately half the samples had higher energy in the mp3 compared to the wav file but it was absurdly close. In frequency terms the two are very close until reaching high frequencies and this is with a fairly old lame encoder, far from the latest.
 
 

 
 
 
So tell me from the below which is wav and which is mp3 

 

 
here are graphs of the bass region 2hz - 401 hz - not much difference at all, i.e mp3 seems to do a decent job of capturing the low end
 

 
Jun 28, 2015 at 4:04 PM Post #15 of 17

nick_charles

Headphoneus Supremus
Joined
Feb 26, 2008
Posts
3,180
Likes
334
  320k shouldn't roll off at 16k - what encoder did you use?

 
an old Lame, 3.98 or close - oops sorry the roll off proper does not start till 19k but the mp3 is starting to diverge noticeably downwards from the wav at about 16.5K
 
16529-90.961349-91.20768-0.24633
16532-92.307854-89.7017672.606087
16535-92.816093-88.2082444.607849
16538-90.405373-86.8604353.544938
16540-90.104271-87.9719012.13237
16543-88.89135-89.728149-0.8368
16546-88.40004-90.094109-1.69407
16548-88.999825-89.823883-0.82406
16551-89.999741-89.4386290.561112
16554-90.479332-89.2667391.212593
16556-92.287819-90.0698322.217987
16559-91.369057-89.8373111.531746
16562-89.71907-88.3176351.401435
16564-89.362602-87.0267792.335823
16567-89.160347-86.3212362.839111
16570-90.080551-87.2920462.788505
16572-91.430321-89.5587081.871613
 
Sadly the above range is a foreign country to me anyway

 
, the actual cliff is at 19.8K 
 

Users who are viewing this thread

Top