How Well Can You Hear Audio Quality?
Jun 3, 2015 at 12:29 PM Post #2 of 22

nntnam

Head-Fier
Joined
Apr 30, 2013
Posts
94
Likes
25
My result using stax 507 & 323s - medium volume: Only one correct (WAV), one 320kbps, the rest of my picks are 128kbps 
confused_face.gif
.
 
Bad news: I have no golden ears.
Good news: I can save a lot of space in my hard disk as I'm gonna covert all of my lossless files into mp3 kbps
k701smile.gif
. They definitely sound the best to my ears.
 
Jun 3, 2015 at 2:21 PM Post #4 of 22

nick_charles

Headphoneus Supremus
Joined
Feb 26, 2008
Posts
3,180
Likes
334
3 right (Neil Young, Jay Z and Mozart) , the other three I picked the 320Kbps - purely guessing I should have gotten 2 right so my performance not significant  I am 56 and everything above 13Khz is a foreign country to me ! but some pretty awful tracks with horrible boomy bass, yeuch ! Easy to do your own by taking a nice clean track and nobbling it !
 
Jun 3, 2015 at 4:43 PM Post #5 of 22

Baxide

100+ Head-Fier
Joined
May 7, 2012
Posts
423
Likes
32
I find many of these audio challenges not much better than trying to pick someone out in a police ID line up. A complex poorly mastered piece might actually sound better after it has been "tidied up" though loss of data bits when converting to mp3 or any of the lossy formats.
 
Jun 3, 2015 at 6:49 PM Post #6 of 22

nick_charles

Headphoneus Supremus
Joined
Feb 26, 2008
Posts
3,180
Likes
334
  I find many of these audio challenges not much better than trying to pick someone out in a police ID line up. A complex poorly mastered piece might actually sound better after it has been "tidied up" though loss of data bits when converting to mp3 or any of the lossy formats.

 
A good digital playback system should perfectly render any arbitrarily complex waveform, lossy codecs (from circa 2005 onwards) should not help as they try to invisibly remove the inaudible masked components, but their design does not hint that they will remove any grunty, if anything they are more likely to add distortion/clipping if they are badly implemented, I've found this a few times on hot tracks where LAME 3.9 something gave rise to audible and measurable clipping when converting a file, though in my experience this is rare.
 
Jun 3, 2015 at 10:25 PM Post #7 of 22

Gowry

100+ Head-Fier
Joined
Jan 9, 2009
Posts
183
Likes
62
A friend at work sent this to me, and I figured someone here would have a thread.
Using a Hugo and LCD-X, I picked out 4 WAV and 2 320Kbps.
It was a very near thing. The 128K didn't have the artifacts I remembered hearing before. I dunno if the MP3 encoders are better or if my DAC is the reason.
 
Jun 3, 2015 at 10:44 PM Post #8 of 22

jcx

Headphoneus Supremus
Joined
Jul 24, 2002
Posts
2,371
Likes
368
I'm sure its quite possible to find source that can't be ABX from the 128k mp3 version if it doesn't have content that excites any of the weaknesses in the lossy compression algorithm
 
during perceptual codec development several "killer samples" - tracks that were difficult for the algorithms, created audible artifacts, were found and passed around and used as standards for further algorithm improvement
 
even 128k mp3 coding improved from its early years
 
Jun 4, 2015 at 2:46 AM Post #9 of 22

Toom

1000+ Head-Fier
Joined
Apr 6, 2015
Posts
1,095
Likes
594
Location
London
  I find many of these audio challenges not much better than trying to pick someone out in a police ID line up. A complex poorly mastered piece might actually sound better after it has been "tidied up" though loss of data bits when converting to mp3 or any of the lossy formats.

 
What do Police ID line ups have to do with this? 
 
Jun 4, 2015 at 4:56 AM Post #11 of 22

WraithApe

Headphoneus Supremus
Joined
Mar 5, 2013
Posts
4,392
Likes
2,283
Location
London, UK
 
during perceptual codec development several "killer samples" - tracks that were difficult for the algorithms, created audible artifacts, were found and passed around and used as standards for further algorithm improvement
 

 
Out of curiosity, do you know what any of these killer samples were? I've read that Tom's Diner, one of the 6 tracks in this test, was used during development of the mp3 codec, but I don't know of any other specific tracks held up as benchmarks for testing the perceptual effectiveness of compression - would be interested to know.
 
Jun 4, 2015 at 11:27 AM Post #12 of 22

arnyk

Repeatedly defended arguments with personal attacks.
Joined
May 30, 2015
Posts
737
Likes
116
   
Out of curiosity, do you know what any of these killer samples were? I've read that Tom's Diner, one of the 6 tracks in this test, was used during development of the mp3 codec, but I don't know of any other specific tracks held up as benchmarks for testing the perceptual effectiveness of compression - would be interested to know.

 
Depends on the coder. The musical selections used in the development of one of the early MPEG coders was detailed in a JAES paper by James Johnston, and did include Solitude Standing from the source you mentioned.
 
Jun 4, 2015 at 2:17 PM Post #13 of 22

sonitus mirus

Headphoneus Supremus
Joined
Mar 23, 2009
Posts
1,781
Likes
612
   
Depends on the coder. The musical selections used in the development of one of the early MPEG coders was detailed in a JAES paper by James Johnston, and did include Solitude Standing from the source you mentioned.

 
I believe that Steve Eddy had contacted JJ within the last month or so with regards to what type of sound characteristics would provide a challenge to the mp3 codec.  SE had quoted JJ's reply, but I don't think I'd be able to locate the comment without some help.
 
Edit: The Google is strong with this one.
 
http://www.head-fi.org/t/486598/testing-audiophile-claims-and-myths/4710#post_11550261
 
Jun 4, 2015 at 3:00 PM Post #14 of 22

peterinvan

100+ Head-Fier
Joined
Jul 23, 2014
Posts
362
Likes
86
Location
Canada
One out of six, using Win7 > Chrome > Meridian Explorer > Tube Amp > Mission V66 tower speakers

Two out of six, using Win7 > Chrome > Meridian Explorer > Modded Fostex T50RPs

Similar results on the Tidal test.

However, comparing HiFi Tidal vs. my own 320Kbs ripped tracks, I can hear the improvement in Tidal's HiFi.
 
Jun 4, 2015 at 3:48 PM Post #15 of 22

sonitus mirus

Headphoneus Supremus
Joined
Mar 23, 2009
Posts
1,781
Likes
612
One out of six, using Win7 > Chrome > Meridian Explorer > Tube Amp > Mission V66 tower speakers

Two out of six, using Win7 > Chrome > Meridian Explorer > Modded Fostex T50RPs

Similar results on the Tidal test.

However, comparing HiFi Tidal vs. my own 320Kbs ripped tracks, I can hear the improvement in Tidal's HiFi.

 
What encoder are you using to rip 320 kbps tracks?  I find no difference in my own ABX tests between Lame 3.99 at -vbr0 using either Exact Audio Copy or dBPoweramp and Tidal's HiFi (FLAC/ALAC) format.
 
Now, it has been established that the Tidal listening test here, http://test.tidalhifi.com/, has an audible and verifiably measurable difference between their own HiFi (FLAC/ALAC) and their lossy AAC 320 kbps files. http://www.head-fi.org/t/743658/tidal-lossless-listening-test-whats-going-on-here/15#post_11077298
 
I can quite easily pass the Tidal listening test 15/15, but I can't pass an ABX between a Tidal lossless test sample and that same file converted with dBPoweramp to mp3.  So, I am able to convert from lossless to lossy with no discernible difference identified through ABX, but somehow Tidal is unable to do the same thing?
 
Edit:
 
I grabbed 2 of the test files from this post, http://www.head-fi.org/t/743658/tidal-lossless-listening-test-whats-going-on-here/15#post_11077558, and could pass an ABX, even with my refrigerator compressor running in the background during the test.  I can take the same lossless file from this test, convert it myself to lossy, and my results are null.  Also, I can take my own lossy version and successfully pass an ABX with the Tidal lossy file.  Try it yourself.  You may have different results.  I don't consider myself to be a particularly critical listener, and I think a well-encoded mp3 or aac is indistinguishable from lossless with the music I've heard.  Though, I can pass an ABX between the Tidal lossless and their lossy files either directly from the test website or using Foobar.
 
foo_abx 2.0.1 report
foobar2000 v1.3.7
2015-06-04 19:02:23
File A: blake_30.flac
SHA1: 470de444f5e02ee225f4234e60b7367ab052c1c9
Gain adjustment: -3.25 dB
File B: blake_30_lossy.flac
SHA1: 9a1f48716a0d04d431c1288e92de4dec721c1b50
Gain adjustment: -2.55 dB
Output:
WASAPI (event) : Speakers (2- USB Modi Device), 24-bit
Crossfading: NO
19:02:23 : Test started.
19:09:25 : 01/01
19:09:53 : 02/02
19:10:19 : 03/03
19:10:54 : 04/04
19:11:41 : 05/05
19:11:56 : 06/06
19:12:41 : 07/07
19:12:52 : 08/08
19:13:19 : 09/09
19:13:45 : 10/10
19:13:45 : Test finished.
 ---------- 
Total: 10/10
Probability that you were guessing: 0.1%
 -- signature -- 
c039abebe6102a987fb1ec62fe11993a238daeba
 
 
If you can't pass the Tidal test, when these are clearly shown to be different, yet you claim to be able to hear a difference with your own converted files, I strongly suggest that you take a close look at whatever process/tools you are currently using to rip a CD and convert to a lossy format.
 

Users who are viewing this thread

Top