Audibility of "typical" Digital Filters in a Hi-Fi Playback

Topic: Audibility of "typical" Digital Filters in a Hi-Fi Playback (Read 331605 times) previous topic - next topic

0 Members and 7 Guests are viewing this topic.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #525 – 2014-11-23 15:53:30

Quote from: amirm on 2014-11-23 15:28:06

I am asking because I post the above but I am accused of cheating by xnor and Ammar.

That is a lie. I said the tests were cheatable.
I said you are determining the files apart by means other than hearing audible differences. You have already admitted this:

Quote from: amirm on 2014-11-22 18:42:49

You say I have claimed to hear audible problems. I have not. I have shown that I can tell files apart.

However, none of this is relevant to the BS paper and the blind tests they performed to obtain their results. As a matter of fact, the BS paper largely dismisses the results of ABX type tests, which would include your online cheatable games versions.
You can't have it both ways Amir.

cheers,

AJ

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #526 – 2014-11-23 16:06:50

Quote from: amirm on 2014-11-23 15:43:27

As to running them again, I am sorry but I won't do it.

I agree, as ABX (specifically unsupervised tests run on Windows computers) is generally poor for these small impairments, especially the psychogenic ones created by "worry"!!!
I'd suggest MUSHRA, but your friend/neighbor within state JJ would be ideal to select, set up and administer such a test/method for you Amir.
Then the whole world would know who Amir Bolt is.
The arguments that Hi-Rez 2ch is an utter $cam would be laid to rest, as would this doctored BS paper.
You hold the keys in your hands for your camp Amir. Victory is possible. What a shame the car won't start.

cheers,

AJ

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #527 – 2014-11-23 16:14:20

Quote from: ajinfla on 2014-11-23 15:39:52

I wonder what sort of non-linearties might appear when a DR 1" Beryllium metal dome is driven that hard with and without 22k band limiting? Might there be some down in the audible bandwidth <20k? When 1' DR domes typically start exhibiting as much as 3db compression at a mere 95db? Hmmm...

I already addressed this Ammar. Here is the explanation again.

The peak SPL measurements are performed using test signals, not the music that the listeners heard. We have the full spectrum of the music that was heard in the paper:

. The total energy hitting the tweeter is pretty small in the ultrasonic region.

I don't know the crossover frequency of the speakers. But taking a guess of a few Khz, the level there averages about 50 db with peaks of about 70 db. Far cry from 95 db.

Quote

Ooops. Oh, that's right, the paper has zero relevant info regarding the tweeters performance within the audible band, at 108db, or anything regarding speakers transparency to the test.

No one listened to full spectrum white noise at 108 db spl. They listened to music. The relevant information is right up there in the measurements.

Quote

Maybe the utter lack of switching software transparency info? The filter "emulations"?

Many other people know that answer here but none are replying to correct your misconception. Anyway, I will explain it to you.

The filters are created in Matlab. Matlab is a math and signal processing program and lets one create such transformations as filter with ease. They read in the source file at 192 Khz, create the filtered versions and then using a GUI front-end they wrote for Matlab, switch between the streams. Matlab is either playing unprocessed source file in memory, or the processed one. The software pipeline through the computer is identical otherwise.

There is no hardware switch box. Just as there is none in Foobar when we compare two files with it.

So the switching fixture is transparent by nature.

The lack of clarity that you speak of symptomatic of not knowing the basics of tools and techniques in this field. That is to be expected. The paper as I keep saying, is written for other industry experts, not lay hobbyists on forums with no signal processing background.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #528 – 2014-11-23 16:24:11

Quote from: ajinfla on 2014-11-23 15:53:30

Quote from: amirm on 2014-11-23 15:28:06
I am asking because I post the above but I am accused of cheating by xnor and Ammar.

That is a lie. I said the tests were cheatable.

Well, this is your chance to clarify your position.

I pulled in the files as presented to me for the tests into the Foobar playlist. I select them, right click and invoke ABX plug-in. I select a segment and listen to A and B to see if I can reliably tell them apart. If I think I can, then I attempt to identify X/Y as being one of them.

If I fail, I go back to pick a different segment. Here is example results:

foo_abx 1.3.4 report
foobar2000 v1.3.2
2014/07/11 06:18:47

File A: C:\Users\Amir\Music\AIX AVS Test files\Mosaic_A2.wav
File B: C:\Users\Amir\Music\AIX AVS Test files\Mosaic_B2.wav

06:18:47 : Test started.
06:19:38 : 00/01 100.0%
06:20:15 : 00/02 100.0%
06:20:47 : 01/03 87.5%
06:21:01 : 01/04 93.8%
06:21:20 : 02/05 81.3%
06:21:32 : 03/06 65.6%
06:21:48 : 04/07 50.0%
06:22:01 : 04/08 63.7%
06:22:15 : 05/09 50.0%
06:22:24 : 05/10 62.3%
06:23:15 : 06/11 50.0% <---- difference found reliably. Note the 100% correct votes from here on.
06:23:27 : 07/12 38.7%
06:23:36 : 08/13 29.1%
06:23:49 : 09/14 21.2%
06:24:02 : 10/15 15.1%
06:24:10 : 11/16 10.5%
06:24:20 : 12/17 7.2%
06:24:27 : 13/18 4.8%
06:24:35 : 14/19 3.2%
06:24:40 : 15/20 2.1%
06:24:46 : 16/21 1.3%
06:24:56 : 17/22 0.8%
06:25:04 : 18/23 0.5%
06:25:13 : 19/24 0.3%
06:25:25 : 20/25 0.2%
06:25:32 : 21/26 0.1%
06:25:38 : 22/27 0.1%
06:25:45 : 23/28 0.0%
06:25:51 : 24/29 0.0%
06:25:58 : 25/30 0.0%
06:26:24 : Test finished.

----------
Total: 25/30 (0.0%)

You can see how I went from not being able to tell the difference, to being able to do so with 100% confidence. Not some total that statistically was valid. But with absolutely correctness in trial after trial.

I hereby represent that the above is all I did. I did not use a spectrum analyzer. I did not edit the log file. I listened until I could identify A/B in double blind ABX testing. That is it. There is no more to the story.

I like to bring all of this to a close Ammar. Please clearly state on the record that you do or do not trust above to be what I have done. If it is not, then please state on what basis/data you are claiming so. Otherwise I hope this is the last time you make an accusation of unethical conduct to me.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #529 – 2014-11-23 16:57:41

Quote from: amirm on 2014-11-23 16:24:11

foo_abx 1.3.4 report
foobar2000 v1.3.2
2014/07/11 06:18:47

File A: C:\Users\Amir\Music\AIX AVS Test files\Mosaic_A2.wav
File B: C:\Users\Amir\Music\AIX AVS Test files\Mosaic_B2.wav

06:18:47 : Test started.
06:19:38 : 00/01 100.0%
06:20:15 : 00/02 100.0%
06:20:47 : 01/03 87.5%
06:21:01 : 01/04 93.8%
06:21:20 : 02/05 81.3%
06:21:32 : 03/06 65.6%
06:21:48 : 04/07 50.0%
06:22:01 : 04/08 63.7%
06:22:15 : 05/09 50.0%
06:22:24 : 05/10 62.3%
06:23:15 : 06/11 50.0% <---- difference found reliably. Note the 100% correct votes from here on.
[... snip ...]
06:26:24 : Test finished.

----------
Total: 25/30 (0.0%)

You can see how I went from not being able to tell the difference, to being able to do so with 100% confidence. Not some total that statistically was valid. But with absolutely correctness in trial after trial.

I hereby represent that the above is all I did. I did not use a spectrum analyzer. I did not edit the log file. I listened until I could identify A/B in double blind ABX testing. That is it. There is no more to the story.
[/color][/b]

You also admit to not hearing an audible difference.
>>> 0 credibility <<<

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #530 – 2014-11-23 16:59:36

So that others, if they so choose, can replicate your work, Amir, would you please provide the approximate time for what you heard as well as describe the phenomena?

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #531 – 2014-11-23 17:36:01

Quote from: amirm on 2014-11-23 15:28:06

Quote from: MLXXX on 2014-11-23 07:22:22
Quote from: amirm on 2014-11-23 05:11:57
Every DBT thrown at me I have taken and passed.

Amirm, I see you passed a comparison of non-linear filtering on this forum in the 20kHz brickwall filtering thread, and a particular comparison of non-aligned sources on another forum (AVS), in its "take 2" thread. I have not seen your attempt at ABXing linear filtering in the 20kHz brickwall thread, or time-aligned files in the AVS take 2 thread (my own samples X2 and Y2).

Given that your hearing is not as responsive to high frequencies as it used to be, it would be quite reasonable for you to decline. However if you wish to display a current prowess, then I would invite you specifically to try ABXing the linear 20kHz brickwall filter example on this forum,¹ and the time-aligned, level aligned, linear filtered for 44.1kHz, On_The_Street_Where_You_Live examples X2 and Y2 on AVS Forum.² If you try, and hear no differences, please advise.

Hi ML. Good to see another familiar face here and a more reasonable one at that.

Turns out I did run one of the tests above. The results were posted in this thread though not that one. David was kind enough to link to it:

Quote from: 2Bdecided on 2014-11-13 17:25:59
Another ABX report of these samples can be found here:
http://www.hydrogenaud.io/forums/index.php...st&p=880594

Here is the results once more:
----------

I gave it a try on the same clip MLXXX had done:

foo_abx 1.3.4 report
foobar2000 v1.3.5
2014/11/13 08:40:05

File A: C:\Users\Amir\Music\HA Forum Tests\limehouse\limehouse_maximum_phase_100.wav
File B: C:\Users\Amir\Music\HA Forum Tests\limehouse\limehouse_reference.wav

08:40:05 : Test started.
08:40:37 : 00/01 100.0%
08:42:19 : 00/02 100.0%
08:43:22 : 00/03 100.0%
08:44:21 : 01/04 93.8% <--- Difference found.
08:45:14 : 02/05 81.3%
08:45:21 : 03/06 65.6%
08:45:34 : 04/07 50.0%
08:45:43 : 05/08 36.3%
08:45:52 : 06/09 25.4%
08:46:00 : 07/10 17.2%
08:46:10 : 08/11 11.3%
08:46:20 : 09/12 7.3%
08:46:29 : 10/13 4.6%
08:46:39 : 11/14 2.9%
08:46:51 : 12/15 1.8%
08:47:00 : 13/16 1.1%
08:47:10 : 14/17 0.6%
08:47:18 : 15/18 0.4%
08:47:26 : 16/19 0.2%
08:47:34 : 17/20 0.1%
08:47:42 : 18/21 0.1%
08:47:49 : 19/22 0.0%
08:47:55 : Test finished.

----------
Total: 19/22 (0.0%)

=========
I also re-ran it with the latest ABX plug-in with signatures:

2014-11-13 09:16:06

File A: limehouse_maximum_phase_100.wav
SHA1: 722dc26db8d4ce666dc03875b2c8d4570d22b521
File B: limehouse_reference.wav
SHA1: e8ad96830d23cad4bba5bf822ce875ae452b9e7c

Output:
DS : Primary Sound Driver

09:16:06 : Test started.
09:16:48 : 01/01
09:16:56 : 02/02
09:17:04 : 03/03
09:17:14 : 04/04
09:17:21 : 05/05
09:17:29 : 06/06
09:17:38 : 07/07
09:17:45 : 08/08
09:17:52 : 09/09
09:18:02 : 10/10
09:18:08 : 11/11
09:18:14 : 12/12
09:18:20 : 13/13
09:18:28 : 14/14
09:18:36 : 15/15
09:18:42 : 16/16
09:18:42 : Test finished.

----------
Total: 16/16
Probability that you were guessing: 0.0%

-- signature --
5b42b06c414b6ba77a3998695bf119a2d57663c0

===========

While I have you here, would you please represent if you cheated in any way in running the above test and getting positive outcome? I am asking because I post the above but I am accused of cheating by xnor and Ammar. More in the next post.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #532 – 2014-11-23 17:49:39

Quote from: xnor on 2014-11-23 16:57:41

You also admit to not hearing an audible difference.

No I have not. I said I heard no audible *problems*. That is *not* the same as audible differences. This is what you asked me:

Quote from: amirm on 2014-11-22 18:42:49

Quote from: xnor on 2014-11-22 17:47:36
Not only did you claim to hear clear audible problems, but run around showing off ABX logs. You say all it comes down to, to hear this 21+ kHz ringing, is training.

Oh? Again, the plot is lost. Here is my comment that you objected to and caused the 12-part replies:

Quote from: amirm on 2014-11-19 18:54:40
Having read the Stuart paper and prior citations, I am starting to think this may be due to time domain impact. It certainly is not frequency domain since I can't hear the ultrasonics. Just as pre-echo is a very audible time domain artifact (although created in frequency domain), maybe these filters act the same way.

You say I have claimed to hear audible problems. I have not. I have shown that I can tell files apart. That is what is required of me: detect lack of transparency.

A problem denotes something specific. ABX testing does not diagnose problems. Its sole job is to see if the tester can tell the files apart by listening. That is what I have done. I listened to the files and found the difference. That is it. I explained every step in my previous post.

Quote

>>> 0 credibility <<<

It is not material what a random poster on a forum thinks of the data in front of them. No one learns anything from that unless you are telling us you are some kind of authority to be believed with no foundation to your accusations. Are you?

I have asked you to contribute constructively to the discussion by running the tests that you say we have must have cheated to pass. You won't. Instead you keep getting personal this way which does nothing but raise the noise floor of the forum.

Since you won't listen to my pleading to put aside your emotions and act objectively, please allow me to deal with this differently since this is what is dominating the discussion now:

Does the larger membership and forum moderators think this is what the discussion should be? That we ask for ABX DBT results as the forum TOS #8 requires but the moment it is present, we call the person a cheater? Is this how you like to be known?

I personally am fine with folks going there. It shows total lack of objectivity and interest in the technical topic. It becomes a referenceable link in the future when we stand on high horse and claim we are all about double blind tests. We are that if it produces negative results. If it produces positive, oh, the person must have cheated. Sorry, no. That is not objectivity. That is biased thinking.

Edit: highlighted above in red.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #533 – 2014-11-23 18:00:43

Quote from: Chu Gai on 2014-11-23 16:59:36

So that others, if they so choose, can replicate your work, Amir, would you please provide the approximate time for what you heard as well as describe the phenomena?

No. As I have mentioned, foobar does not save the segments markers and I have no record of them as such.

I did provide some data though in one of the tests: http://www.whatsbestforum.com/showthread.p...ll=1#post277513

Quote

Not quite the topic of the thread but nevertheless part of similar arguments, here is a comparison of Arny's 24/96 file downsampled to 16/44.1 compared to 320 kbps MP3 of the same:

=================

foo_abx 1.3.4 report
foobar2000 v1.3.2
2014/07/19 19:45:33

File A: C:\Users\Amir\Music\Arnys Filter Test\keys jangling 16 44.wav
File B: C:\Users\Amir\Music\Arnys Filter Test\keys jangling 16 44_01.mp3

19:45:33 : Test started.
19:46:21 : 01/01 50.0%
19:46:35 : 02/02 25.0%
19:46:49 : 02/03 50.0% << dog barked in my ear wanting to go out
19:47:03 : 03/04 31.3%
19:47:13 : 04/05 18.8%
19:47:27 : 05/06 10.9%
19:47:38 : 06/07 6.3%
19:47:46 : 07/08 3.5%
19:48:01 : 08/09 2.0%
19:48:19 : 09/10 1.1%
19:48:31 : 10/11 0.6%
19:48:45 : 11/12 0.3%
19:48:58 : 12/13 0.2%
19:49:11 : 13/14 0.1%
19:49:28 : 14/15 0.0%
19:49:52 : 15/16 0.0%
19:49:56 : Test finished.

----------
Total: 15/16 (0.0%)

I just selected the beginning of the file and the difference was very clear to my ears.

So let's have you all run the same test and see what you hear. Please don't keep asking me to explain to you what a Quince tastes like. I can't. You need to eat one (preferably a ripe one!). Better yet, smell one.

Note that as I have quoted multiple times, we do have such data for Stuart's test:

It was reported that filtering gave "softer edges" to
the instruments, and "softer leading edges" to musi-
cal features with abrupt onsets or changes. Echoes,
when audible, were identied as being aected the
most clearly by the ltering. It was felt that some
of the louder passages of the recording were less ag-
gressive after ltering, and that the inner voices (sec-
ond violin and viola) had "a nasal quality". Over-
all, the ltered recording gave a "smaller and
atter auditory image", and specically the physical space
around the quartet seemed smaller.

Listeners described that quantization gave a "rough-
ness" or "edginess" to the tone of the instruments,
and that quantization had a signicant impact on
decay, particularly after homophonic chords, where
"decay was sustained louder for longer and then died
suddenly". This could be an eect of quantization
distortion; it is interesting that this was audible even
in a 24-bit system, and is consistent with the hy-
potheses of Stuart [29] that 16 bits are not sucient
for inaudible quantization.

Did this do anything for you? I suspect the answer is no .

Just take the test for heaven's sake. Why is there so much hesitation to do so? Take it. Listen carefully. Try hard. You know others did it. Use that as motivation to find the differences and become a more critical listener. This is what others who took the test have gained.

Start with the easier MP3 case above. If you can't tell the difference at 320, try it at 256. Keep going lower until you do. Then work backward to higher bit rates and focus on the same differences you heard before.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #534 – 2014-11-23 18:03:29

Quote from: amirm on 2014-11-23 15:28:06

Here is the results once more:
----------

I gave it a try on the same clip MLXXX had done:

foo_abx 1.3.4 report
foobar2000 v1.3.5
2014/11/13 08:40:05

File A: C:\Users\Amir\Music\HA Forum Tests\limehouse\limehouse_maximum_phase_100.wav
File B: C:\Users\Amir\Music\HA Forum Tests\limehouse\limehouse_reference.wav

...
===========

While I have you here, would you please represent if you cheated in any way in running the above test and getting positive outcome? I am asking because I post the above but I am accused of cheating by xnor and Ammar.

Well with the maximum phase file compared with the reference file, I did hear a slight dullness, back in January 2009. At the time my equipment was fairly basic. I used pc motherboard high definition sound, and medium quality hi-fi speakers. (These days if attempting such an ABX I'd use my Xonar sound card, and Sennheiser HD 800 headphones.) I certainly didn't cheat. I simply performed the ABX with foobar, once I felt confident that I could hear a difference.

A maximum phase filter can delay frequencies extending well down into the audible range. I have found that the timbre of a tone with harmonics can change if one of the harmonics is altered in phase relative to the fundamental. [As a simple practical test, if I use Audacity at a sample rate of 96kHz to create a track with a 3kHz sinewave at a level of 0.5, and another track with a 9kHz sinewave at a level of 0.1, the addition of the 9kHz track alters the character of the sound. If I then delay the 9kHz tone by 90 degrees and play both tracks again, I find the sound has a slightly different character again, compared with the 9kHz tone being in phase.]

I was really more interested in your trying 2Bdecided's linear phase file _50.wav (as I mention below). I recall that I was unable to discern any audible difference with that file, in January 2009.

Quote from: amirm on 2014-11-23 15:43:27

Quote from: MLXXX on 2014-11-23 07:22:22
Given that your hearing is not as responsive to high frequencies as it used to be, it would be quite reasonable for you to decline. However if you wish to display a current prowess, then I would invite you specifically to try ABXing the linear 20kHz brickwall filter example on this forum,¹ and the time-aligned, level aligned, linear filtered for 44.1kHz, On_The_Street_Where_You_Live examples X2 and Y2 on AVS Forum.² If you try, and hear no differences, please advise.

I addressed your first request in my last post.

You may have misunderstood my first request. I was not asking you to reproduce an ABX log for 2Bdecided's maximum phase file, as I had already seen an ABX log by you for that file. I was asking you to try his linear ("50") phase file, a much more difficult challenge, given that a linear phase file for 22.05 kHz should not affect the phase of frequencies in the audible range. I felt this request quite relevant, in light of your suggestion that the [linear] filtering for the BS paper might have been audible, on account of its effects in the time domain.

Quote from: amirm on 2014-11-23 15:43:27

On running your set of converted AVS files, as you know, I did give them a try by downloading the 150 (?) megabyte package:

Quote
First, thanks so much for creating and uploading these files. This is what I call constructive discussion .

Something is very puzzling or drastically wrong though if I am understanding you right. I compared X and Y against each other in foobar. I immediately heard a difference at the start of the tracks. [color="#FF0000"]One has tons more noise added to it. [/color] You can hear it both before the music starts and during. The noise levels are very high and not consistent with proper dither.

I pulled the files into Audition and played them with the same issue. I zoomed into the start of the "Y" file and it indeed has noise that is elevated to -75 dbFS (~12 bits of signal to noise ratio). The file shows the 22 Khz cut off so it is the Audacity converted one. This is really, really broken if that is the conversion that Audacity performed. Would you please verify your workflow and confirm what I am saying here? Because if correct, it means Audacity signal processing is worse than junk. And the generated noise will really screw up any "null" type tests as it will dwarf signal dependent differences.

Members here should notice that I did not provide any ABX results. Would have been an easy win. But not an ethical one. So I didn't and instead let you know what was broken in the test.

As to running them again, I am sorry but I won't do it. The mountain of ABX results I have provided has meant nothing to the objectors here. This is tedious work and something I don't enjoy doing. But let's ask other people to do it and post their results.

I didn't provide an extended response on AVS forum at the time when the criticism was first made, but after having had time to analyse the files I am sorry but I disagree. A null test with the files I initially provided is in fact quite successful at a moderate listening volume. With moderate gain, the dither cannot be heard, not for my hearing anyway! The frequency components peaked at -104dB at around 9kHz, as can be seen in the first graph [after accounting for the 30dB amplification] in the attachment to AVS forum post #564 of the AVS Forum thread AVS/AIX High-Resolution Audio Test: Take 2. I do not see noise at -75dB in any frequency band in the first graph for the first second (after allowing for the 30dB amplification I performed before running the frequency analysis). I would concede that applying 30dB amplification to low level quantisation/dither noise for the purpose of a frequency analysis graph might create some degree of anomaly in accuracy of the dB levels, as the noise is of such a low level, spread over a very wide range of frequencies.

On reflection, I don't think the conversion was "broken", merely that the dither that the old version of Audacity had created was somewhat noisy. However, as it was noisier than usual practice, and in view of the complaints, by yourself, and others, I went to the trouble of creating replacement files X2 and Y2, using SoX, set for a slightly modified TPDF dither.

I note that one person on AVS (chip_mk) did comment on my files X2 and Y2. He stated that they were subjectively more similar to each other than the AVS Forum A2 and B2 files (for On_The_Street_Where_You_Live). Despite the lack of corroboration of the AVS forum member's claims by way of ABX software, I took this subjective comment as a vindication of my efforts. Certainly for my ears the AVS files had a difference in "tone" in addition to the misalignment of approximately 10mS. I think these factors were enough to invalidate the AVS Forum exercise, i.e. it was too easy to tell the files apart (as I had done with a formal ABX). Extraneous factors had crept in, firstly the lack of time alignment, and secondly some tonal difference. I am unable to explain how the tonal difference arose. Why AVS Forum could not simply have provided time-aligned files capable of being audibly nulled at a moderate listening gain (by inverting one of the files) is a little surprising. I guess it was due in part to lack of familiarity with this type of exercise, and the importance of matching the files. Not only were the files not time aligned, but even when time aligned and with one file inverted, they would not null to silence at a moderate gain setting.

I agree with your comment that ABX testing is tedious work. It can be even more tedious if differences are very subtle. I take on the challenge only very occasionally myself. I make a record of the character of the difference for my hearing.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #535 – 2014-11-23 18:43:22

Quote from: amirm on 2014-11-23 16:14:20

Quote from: ajinfla on 2014-11-23 15:39:52
Ooops. Oh, that's right, the paper has zero relevant info regarding the tweeters performance within the audible band, at 108db, or anything regarding [!--sizeo:3--][span style=\"font-size:12pt;line-height:100%\"][!--/sizeo--]speakers[/size] transparency to the test.
That's too bad.
The peak SPL measurements are performed using test signals, not the music that the listeners heard.
Quote
For a system gain of 75 dB, the loudest peak [!--sizeo:4--][span style=\"font-size:14pt;line-height:100%\"][!--/sizeo--]passage[/size] was measured as 102 dB SPL at the listening position, somewhat lower than the level we would expect from a live performance at a distance of 3 m.
The irrelevant information is right up there in the measurements.

Quote

1pas·sage
noun \ˈpa-sij\

b : a phrase or [!--sizeo:3--][span style=\"font-size:12pt;line-height:100%\"][!--/sizeo--]short section of a musical composition[/size]

Sorry Amir, this loudspeaker stuff is way over your head. Huge problem when hobbyists try to analyze deliberately cryptic professional work.

Quote from: amirm on 2014-11-23 16:14:20

Quote from: ajinfla on 2014-11-23 15:39:52
Maybe the utter lack of switching [!--sizeo:3--][span style=\"font-size:12pt;line-height:100%\"][!--/sizeo--]software[/size] transparency info?
There is no hardware switch box. Just as there is none in Foobar when we compare two files with it.
So the switching fixture is transparent "by nature".

"Worry" syndrome obviously creates serious reading comprehension problems.

Quote from: amirm on 2014-11-23 16:14:20

The lack of clarity that ~~you~~ I speak of with is symptomatic of not knowing the basics of tools and techniques in this field. That is to be expected. The paper as I keep saying, is written for other industry experts, not lay hobbyists on forums with no signal processing background.

Agreed.

cheers,

AJ

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #536 – 2014-11-23 18:46:47

Quote from: amirm on 2014-11-23 16:24:11

I hereby represent that the above is all I did.

Uh huh.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #537 – 2014-11-23 18:57:13

Quote from: amirm on 2014-11-23 17:49:39

No I have not. [...]

I've already dealt with your excuses. 0 credibility.

Quote from: amirm on 2014-11-23 17:49:39

It is not material what a random poster on a forum thinks of the data in front of them. No one learns anything from that unless you are telling us you are some kind of authority to be believed with no foundation to your accusations. Are you?

1) Exactly.
2) Another fallacy. See 0 credibility.

Quote from: amirm on 2014-11-23 17:49:39

I have asked you to contribute constructively to the discussion by running the tests that you say we have must have cheated to pass. You won't. Instead you keep getting personal this way which does nothing but raise the noise floor of the forum.

Stop posting noise and the noise floor will go down. See 0 credibility.

Quote from: amirm on 2014-11-23 17:49:39

Since you won't listen to my pleading to put aside your emotions and act objectively, please allow me to deal with this differently since this is what is dominating the discussion now:

This is based objectively on what you posted in the past: 0 credibility.

Quote from: amirm on 2014-11-23 17:49:39

Does the larger membership and forum moderators think this is what the discussion should be? That we ask for ABX DBT results as the forum TOS #8 requires but the moment it is present, we call the person a cheater? Is this how you like to be known?

I've already dealt with your fallacious arguments and deliberate obtuseness. See 0 credibility and posts before that.

Quote from: amirm on 2014-11-23 17:49:39

I personally am fine with folks going there. It shows total lack of objectivity and interest in the technical topic. It becomes a referenceable link in the future when we stand on high horse and claim we are all about double blind tests. We are that if it produces negative results. If it produces positive, oh, the person must have cheated. Sorry, no. That is not objectivity. That is biased thinking.

I have a referenceable link for your nonsense: 0 credibility.

Future posts of nonsense will be condensed into the most funny line to spare other readers of risking to read the same nonsense twice accidentally.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #538 – 2014-11-23 19:49:20

Adding to my previous post about ringing.

Here's what happens if you re-apply a less steep 44.1 kHz filter than the one used in the paper:

red = 21.591 kHz passband edge, 22.05 kHz stopband
blue = red filtered with a less steep filter that has a cutoff point of roughly 21 kHz, same stopband

Visualization is done using a high oversampling factor and:

Code: [Select]

20*log10(abs(filter))

For all I care, those who produce the music may use as steep filters as they like, because I'm resampling during playback anyway.
Again, this is without imaging trickery that Meridian actually makes use of in their products... (which distances their claim of "real-world filters" even more from reality).

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #539 – 2014-11-23 19:53:28

Quote from: amirm on 2014-11-23 17:49:39

Since you won't listen to my pleading to put aside your emotions and act objectively, please allow me to deal with this differently since this is what is dominating the discussion now:

Does the larger membership and forum moderators think this is what the discussion should be? That we ask for ABX DBT results as the forum TOS #8 requires but the moment it is present, we call the person a cheater? Is this how you like to be known?

I am only an infrequent contributor to this forum, and lack familiarity with personalities behind a number of the member names, but I have found the level of vitriol in this thread unusually high.

My reaction to an unexpectedly good ABX result would be firstly to query the equipment and the experimental technique. And then if reasonably feasible, attempt to replicate the test result with help from those who claim to have heard a difference (see below).

Quote from: Chu Gai on 2014-11-23 16:59:36

So that others, if they so choose, can replicate your work, Amir, would you please provide the approximate time for what you heard as well as describe the phenomena?

Yes, if the subjective difference in sound is at its most obvious in only selected segments of the files under comparison, it can be very useful to identify where one or more of those segments are. It can help others do their "fishing" at a promising spot! (Sometimes, though, a difference is pretty much continual.)

In any case, it can be very useful to state the subjective character of the difference(s). This can helps others in their attempts at hearing. Rarely, a difference could be so minor and fleeting as to defy characterisation, but even describing it in that vague way could conceivably be of some help.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #540 – 2014-11-23 20:08:28

MLXXX, let me save you some time:

After several pages of painful back and forth:

Quote from: amirm on 2014-11-22 18:42:49

Quote
But you still haven't even given a hint at what specifically you hear.

And I am not going to.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #541 – 2014-11-23 21:26:37

Quote from: xnor on 2014-11-22 21:38:07

Quote from: amirm on 2014-11-22 18:42:49
9. Some Conclusions
Using admittedly exaggerated filter designs

Just stop there amirm. You don't see a problem here?
Also, the cited paper is largely outdated. Over 30 years ago the filters in A/D/A converters looked very different. You know that but you cherry-pick (another fallacy btw).

Oh the "it is too old argument." Arny does that once in a while.

It must be that human evolution in 30 years that make the results invalid. We must hear differently now than then.
Or maybe it is because someone discovered digital filtering in the ensuing 30 years. Digital audio must be a lot younger than we think.

Fortunately for us neither is the case. The paper is referenced in Stuart's. You have not read it. It is not a review of A/D and D/A converters as to be too old. No.

The paper explains how as we went from analog to digital, we introduced distortions that did not exist in analog domain. And that while the general public may not care, professional listeners should.

This is the opening salvo:

1.3 Simple experiments and perceptible effects

A good starting point is to base on experiments which, although
seemingly safe and unobjectionable, still lead to clearly perceptible
effects. They make it possible to identify some of the
reasons for the lack of fidelity in conversion, and to derive
design criteria from them. Fortunately, two such experiments are
known, one involving a seemingly ideal digital filter of absolutely
flat characteristics and linear phase, the other involving
very steep digital filters (again with linear phase). The next
chapter will describe the effects which were encountered with an
"ideal" filter, the explanations which were found for them, and
techniques for eliminating them. The second set of experiments,
which is based on the difference between "gentle" and "brickwall"
digital filters (all of them with linear phase) will be described
at the end of the paper.

This is the reference given right when the FIR filter discussion occurs in Stuart's paper. And you call that cherry picking? We are discussing Stuart paper. The references given there are 100% proper to cite. I didn't go rummage through some archive and pick some random blog to cite. This is work presented by luminaries of the industry and is completely relevant to the current discussion.

I also provided much more recent paper from JJ. Here is that again from his 2006 presentation:
. 30 years later the problem is still acknowledged. And with it being mentioned in Stuart's paper, it still won the peer review award.

Is it true? I don't know. We know it is something to think about. That is all that these references were meant to back. You don't want to think about it, don't.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #542 – 2014-11-23 21:46:14

Quote from: MLXXX on 2014-11-23 18:03:29

I was really more interested in your trying 2Bdecided's linear phase file _50.wav (as I mention below). I recall that I was unable to discern any audible difference with that file, in January 2009.

Oh sorry. I was wondering why at the same time you were saying I took the test and I didn't .

I did very quick try and these are the results for linear phase:

=======================
foo_abx 2.0 beta 4 report
foobar2000 v1.3.5
2014-11-23 13:38:11

File A: limehouse_linear_phase_050.wav
SHA1: 661058f46dfb7de9fd2687344ece857f0ae1531a
File B: limehouse_reference.wav
SHA1: e8ad96830d23cad4bba5bf822ce875ae452b9e7c

Output:
DS : Primary Sound Driver

13:38:11 : Test started.
13:38:31 : 01/01
13:38:39 : 02/02
13:38:48 : 02/03
13:38:58 : 03/04
13:39:06 : 04/05
13:39:18 : 05/06
13:39:24 : 06/07
13:39:31 : 07/08
13:39:40 : 07/09
13:39:48 : 08/10
13:39:56 : 09/11
13:40:03 : 10/12
13:40:09 : 11/13
13:40:17 : 12/14
13:40:28 : 13/15
13:40:36 : 14/16
13:40:36 : Test finished.

----------
Total: 14/16
Probability that you were guessing: 0.2%

-- signature --
b28120a9d23f5f5e1706c1bdd66b655564163639

I distinctly remember losing concentration the couple of times above. But since the results show 99.8% non-chance outcome, I figured it is good enough to post and not suffer any more .

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #543 – 2014-11-23 23:43:31

Quote from: MLXXX on 2014-11-23 18:03:29

I didn't provide an extended response on AVS forum at the time when the criticism was first made, but after having had time to analyse the files I am sorry but I disagree. A null test with the files I initially provided is in fact quite successful at a moderate listening volume.

My impression of the noise is far higher than you and mzil. Maybe we are subjectively different when it comes to perception of it. At the start of the track it sounded like a white noise generator next to your ear while your stereo played .

Quote

On reflection, I don't think the conversion was "broken", merely that the dither that the old version of Audacity had created was somewhat noisy.

That was my point. Unskilled people go and look up signal processing algorithms and implement them. Without critical listening skills, they think the job is done and ship it. This yet again demonstrates that there are two types of listeners: those that can hear non-linear artifacts and those who can't.

Speaking of that, I downloaded your latest tracks. At 50.9 to 51.1 I thought the "s" in street sounded distinctly different. There was more lisping in one than the other. In the trial mode I was able to consistently tell the difference for a good sequence of trials. But when I ran with the test without feedback, I think I got down to 30% probably of chance or some such thing. I am just not motivated to try again and don't remember where the critical segment was in my original testing. So if you like to declare this a loss for me, you can . I am just too lazy to try harder and see if I can pass it.

Quote

Despite the lack of corroboration of the AVS forum member's claims by way of ABX software, I took this subjective comment as a vindication of my efforts. Certainly for my ears the AVS files had a difference in "tone" in addition to the misalignment of approximately 10mS. I think these factors were enough to invalidate the AVS Forum exercise, i.e. it was too easy to tell the files apart (as I had done with a formal ABX).

Yet, hardly anyone has managed to pass that test, with ease or not!

Ultimately what I like people to take to the bank is just that: that we don't hear the same.

The data is 100% compelling in that regard. And very easy to verify. Anyone who disagrees should try to duplicate our results. They can try to find the flaws just the same. But I suspect with full knowledge of all of that, they don't have critical enough listening skills to find and hear the difference in DBT ABX tool.

And given that proof, you cannot extrapolate from your hearing to the rest of the people. Just because you don't hear the difference doesn't mean it is inaudible and sufficient justification for you to go around and accuse people are being wrong to observe otherwise.

Quote

I am unable to explain how the tonal difference arose. Why AVS Forum could not simply have provided time-aligned files capable of being audibly nulled at a moderate listening gain (by inverting one of the files) is a little surprising. I guess it was due in part to lack of familiarity with this type of exercise, and the importance of matching the files. Not only were the files not time aligned, but even when time aligned and with one file inverted, they would not null to silence at a moderate gain setting.

Actually they duplicated a real-life situation. Took a professional audio workstation tool, Sonic Solutions, and converted the files to 16/44.1. Precisely how real music is produced. That in double blind tests we could tell the difference it means that what people observe in the wild can very much be true. That transparency is not there.

It matters not that the timing difference may be the reason. What matters in this context is that we are wrong. We are wrong to say they are imagining things when they compare the high res to 16/44.1. We are making idealized assumptions that are not true in reality.

So I think it is good that we discovered how professional resamplers are not the animals we think they are. There is no visibility into their design as I have mentioned. I imagine hardly any music is produced using Sox resampler. Engineers use the professional tools. And if those tools produce non-transparent results, then we should get the masters and not be subject to this.

These are 100% logical and defensible conclusions we can draw. Anyone who doesn't want to go there has fingers in the ears and in denial.

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #544 – 2014-11-24 00:05:25

Quote from: amirm on 2014-11-23 21:26:37

Oh the "it is too old argument." Arny does that once in a while.

It must be that human evolution in 30 years that make the results invalid. We must hear differently now than then.
Or maybe it is because someone discovered digital filtering in the ensuing 30 years. Digital audio must be a lot younger than we think.

Amir, the master of straw man arguments and red herrings. See 0 credibility.

I've asked you this:

Quote from: xnor on 2014-11-22 21:38:07

But please show us where it demonstrates the audibility of a linear-phase lowpass with a transition band of roughly 2 kHz (that's what we find in modern devices), with a cutoff frequency above ~21 kHz.

This would be a worst case scenario that still matches reality, while the paper uses a transition band of about 450 Hz. No, I'm not saying that is audible normally either, save your straw man arguments (at the risk of bursting).
But even if the files were filtered with ridiculously small bandwidth, we could still filter with a more sane filter and remove the characteristics of the bad filter.

But, yes, you absolutely do cherry-pick whatever fits your agenda, see 0 credibility.
You know that you can make allpass filters, cables, ... or almost any other device or component exhibit audible differences, you just need to ridiculously overdo it. And no honest person would answer a question what that audible difference is with: "I'm not going to tell you", see 0 credibility.
But that is what your argument is now based on.

Quote from: amirm on 2014-11-23 21:26:37

This is the reference given right when the FIR filter discussion occurs in Stuart's paper. And you call that cherry picking?

Oh amir, playing deliberately obtuse again, are we?
"admittedly exaggerated filter designs"

Can you not even accept the words you quote? (Granted, they were not highlighted by that stupid annoying red color you seem to love to use to show us what you cherry-pick.) But then again, you don't even seem to accept that all the filter does is ring at over 21 kHz...

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #545 – 2014-11-24 00:55:33

Quote from: MLXXX on 2014-11-23 19:53:28

My reaction to an unexpectedly good ABX result would be firstly to query the equipment and the experimental technique.

My reaction would be, let's see you pass a supervised, administered listening test, not one sitting unsupervised with a Windows computer. A guy named JJ lives in WA...
30 years of zero audible evidence, now suddenly numerous "passes" of online ABX games on Windows computers, unsupervised in 2014....with a bit of history in between.

And now back to our scheduled programming: "Audibility? of Atypical Digital Filters in an unconfirmed transparency Hi-Fi Playback System, blaring near maximum volume in an iso-ward, without using low specificity ABX".

cheers,

AJ

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #546 – 2014-11-24 00:58:30

Quote from: xnor on 2014-11-24 00:05:25

Amir, the master of straw man arguments and red herrings.

Straman this, red herring that. I have never, ever heard anyone utter those phrases in real life or technical conferences. You would be laughed at for using those high-school debating terms.

Through our bullying tactics, we have made up these rules of discussion. Oh that piece of data hurts? Let me just claim it is "straw man." If that doesn't work, then it must be red herring. If that doesn't work it is appeal to authority. If that doesn't work it is excluded middle or whatever that nonsense is called.

Leave those things to someone who won't call you on it as a substitute for proper technical answer. You clearly demonstrate your lack of knowledge when you use these phrases.

Quote

And no honest person would answer a question what that audible difference is with: "I'm not going to tell you"

Do people have a different definition of honesty in Australia? I have been to your beautiful country and don't recall people speaking differently.

"I am not going to tell you" is just that: a choice. Ton of data has been shared. Research report being discussed. Countless tests by me and others across wide range of test conditions. You have not accepted a single one. Mr. Xnor is righ, and the rest of the world include the experts who wrote the paper in question and others who reviewed it, wrong. Right. More demands for data is made. As if giving them more would make them more civil or interested in knowledge.

I ask you to repeat our listening tests. You know, double blind ABX tests. You refuse. Your refuse! If I were you I would go and change my alias and come back as a different person after this debacle. It is that embarrassing.

You guys talk big about blind tests. Demand it left and right. Put it in the forum TOS. But heaven forbid for you to be in the hot seat to run them. Oh no. The other party is accused of being a cheater, and demand neverending additional data. Anything but running the test yourself.

With straight face you walk around proud of this. Oh I am going to put it in my signature that I won't run any blind test to show whether I can or cannot tell the difference that others have reported beside Amir. Go ahead, I love to see my name in all of your posts. Someone will do a search and land on this thread .

You talk big about science. But you have no idea what the AES organization is, or the provenance of papers published within. On page 1 it is declared to not be peer reviewed. I correct that. It is repeated again by Arny. I correct it one more time. And have to do it a third time at the end of this thread. Is it not clear that the vocal few here don't really read or understand audio literature?

Buying an AES paper? Nah, not going to happen. That costs a few dollars. Stomping your feet in posts is free. Never mind that you learn nothing about audio by doing that. You are going to do it. If it doesn't work, there is always "straw man" this, and "red herring that."

Let me cut through this:

The conclusions are:

1. We don't all perceive non-linear distortions the same. This is proven beyond any doubt whatsoever. Anyone who wants to challenge this should show up passing the same tests. Therefor you cannot extrapolate your hearing as your belief as to what is or is not audible.

2. Our demand of a double blind ABX test is an empty bluf. Should the person pass it under all conditions we define, we can still call them a cheater and do away with the results. We refused to run them ourselves.

3. We now have award winning peer-reviewed paper showing that to better than 95% statistical confidence, filters have audible effects and so does dither/truncation. It is not 100% proof but it is heck of a lot better than anonymous members stomping their feet and singing "liar liar, pants on fire."

There. Live with these .

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #547 – 2014-11-24 01:23:58

Quote from: ajinfla on 2014-11-24 00:55:33

Quote from: MLXXX on 2014-11-23 19:53:28
My reaction to an unexpectedly good ABX result would be firstly to query the equipment and the experimental technique.

My reaction would be, let's see you pass a supervised, administered listening test, not one sitting unsupervised with a Windows computer. A guy named JJ lives in WA...

Don't think MLXXX lives here but whatever.

This is the forum TOS:

8. All members that put forth a statement concerning subjective sound quality, must -- to the best of their ability -- provide objective support for their claims. Acceptable means of support are double blind listening tests (ABX or ABC/HR) demonstrating that the member can discern a difference perceptually, together with a test sample to allow others to reproduce their findings. Graphs, non-blind listening tests, waveform difference comparisons, and so on, are not acceptable means of providing support.

The only test of validity is providing samples for others to reproduce it. That is it. You guys are choosing to not try to do that. There is no other recourse in the forum TOS. Making up new rules about live witnesses and such are as silly as you using $230/foot speaker cable at high end audio shows Ammar.

Quote

30 years of zero audible evidence, now suddenly numerous "passes" of online ABX games on Windows computers, unsupervised in 2014....

The answer is simple: no one had presented a test to run until now. Scott/Mark did so on AVS as did Arny. Arny said no one had ever passed his test and made it into a challenge to do so. He even went as far as saying people could not tell the difference down to 32 Khz. So I gave it a try. And here we are.

Let me share you an interesting tidbit on AVS. Frank was one of my detractors here. He reported this on Arny's test, trying to demonstrate it could not be passed:

============
First ABX results with Jingling keys:

foo_abx 1.3.4 report
foobar2000 v1.3.2
2014/07/09 14:17:03
File A: \\diskstationone\music\jitter\test\keys jangling band resolution limited 3216 2496.wav
File B: \\diskstationone\music\jitter\test\keys jangling full band 2496.wav
14:17:03 : Test started.
17:07:14 : 00/01 100.0%
17:07:41 : 00/02 100.0%
17:08:23 : 00/03 100.0%
17:09:27 : 00/04 100.0%
17:09:45 : Test finished.
----------
Total: 0/4 (100.0%)

My upper hearing range is about 14kHz. I doubt further training will do much good.
====

So total failure. As it happens, I too was running the test in parallel and reported my results:

32 Khz versus 96 Khz
=================================
foo_abx 1.3.4 report
foobar2000 v1.3.2
2014/07/09 06:10:07

File A: C:\Users\Amir\Music\Arnys Filter Test\keys jangling band resolution limited 3216 2496.wav
File B: C:\Users\Amir\Music\Arnys Filter Test\keys jangling full band 2496.wav

06:10:07 : Test started.
06:10:38 : 01/01 50.0%
06:10:50 : 02/02 25.0%
06:11:07 : 03/03 12.5%
06:11:23 : 04/04 6.3%
06:11:36 : 05/05 3.1%
06:12:00 : 06/06 1.6%
06:12:14 : 07/07 0.8%
06:12:26 : 08/08 0.4%
06:12:38 : 09/09 0.2%
06:12:49 : 10/10 0.1%
06:13:00 : 11/11 0.0%
06:13:23 : 12/12 0.0%
06:13:42 : 13/13 0.0%
06:13:48 : Test finished.

----------
Total: 13/13 (0.0%)

A remarkable thing happened. Once he knew this was a test that could be passed, he tries again:

======

A and B along with X and Y I listen from slightly louder 'sparkle' in the hirez file. (Difficult to dicribe the sound)

foo_abx 1.3.4 report
foobar2000 v1.3.2
2014/07/09 20:49:23
File A: \\diskstationone\music\jitter\test\keys jangling band resolution limited 3216 2496.wav
File B: \\diskstationone\music\jitter\test\keys jangling full band 2496.wav
20:49:23 : Test started.
20:51:10 : 01/01 50.0%
20:52:25 : 02/02 25.0%
20:52:46 : 03/03 12.5%
20:53:12 : 04/04 6.3%
20:53:58 : 05/05 3.1%
20:54:20 : 06/06 1.6%
20:54:47 : 07/07 0.8%
20:55:25 : 08/08 0.4%
20:55:49 : 09/09 0.2%
20:56:16 : 10/10 0.1%
20:56:44 : 11/11 0.0%
20:57:15 : 12/12 0.0%
20:57:23 : Test finished.
----------
Total: 12/12 (0.0%)

======

Completely positive identification! Here is a person who says he can't hear past 14 Khz yet he could tell a resample that limit the response to 16 Khz (32 Khz sampling). Xnor, there is your description of the difference. Happy now? I suspect not .

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #548 – 2014-11-24 01:29:54

Quote from: amirm on 2014-11-24 00:58:30

Straman this, red herring that. I have never, ever heard anyone utter those phrases in real life

Of course not, those are taught in Logic classes. Still not too late for you too take one.

Quote from: amirm on 2014-11-24 00:58:30

Research report being discussed. Countless tests by me and others across wide range of test conditions.

Right. You have utterly subverted the discussion about the BS paper/results by conflating it with cheatable/flawed unsupervised online ABX games. Just as quoted above.
Your ABX results are absolutely unrelated to the BS paper/results.
You should start a separate thread on your new methods for passing unsupervised ABX tests on Windows computers. Make that the title if you wish, maybe add: "A follow up to the 2009 +/- 10% volume method for identifying DACs"
It ought to be a doozy.
In the meanwhile, how about addressing the issues raised here about the BS paper and stop polluting the thread with your ABX conflations.

cheers,

AJ

Audibility of "typical" Digital Filters in a Hi-Fi Playback

Reply #549 – 2014-11-24 01:48:03

Quote from: amirm on 2014-11-24 01:23:58

Making up new rules about live witnesses and such are as silly as you using $230/foot speaker cable at high end audio shows Ammar.

Supervision is to ease my type "worries", the cables, yours.
Take that Logic class Amir.

Quote from: amirm on 2014-11-24 01:23:58

The answer is simple: no one had presented a test to run until now.

Baloney. You claim to have been running your own filter artifacts test regiment since 2001

Quote

Originally Posted by amirm
Thanks for creating the thread. I must say though, I feel like I am on trial or something . We are all here for a fun exchange. If what I present doesn't meet some high bar, so be it....

The comparison I performed was using a Mark Levinson No360S against the on-board DACs in five to six DVD-A and SACD players, all playing the same time sync'ed CD. In other words, I would listen to the analog output of the player while its digital output would feed the ML DAC. All front panel lights were turned off in addition to video circuits (yes, all of that made a difference in fidelity).

The two sources were fed to the dual inputs of a Stax "earspeaker" electrostatic headphone amp. If you are not familiar with Stax, you can read learn more about them here: http://www.stax.co.jp/Export/ExportProducts.html. I have three of their units and results are consistent across the board although the highest end unit does make the job a bit easier. Using headphones allowed me to completely eliminate the room and take advantage of the amazing transparency of these headphones to listen for the slightest differences. To latter point, I would often listen to material at levels well above what I would use for listening to music, allowing me to hear detail that would otherwise be lost.

I then picked material that made it easier to detect differences between DACs. I am not going to disclose what constitutes such content. Without such material, the job can range from difficult to impossible. One has to know what could be damaged by a DAC and then use music that has such content. To give you an example, when you compress music, it is the transients that suffer. So something like guitar music is much more revealing than say, violin as the latter is much more harmonic than the sharp impulses of a guitar. Voices play the same role. None of these are useful for testing DACs though so don’t use that as a hint to the question posed . You can’t test the cornering of a car if you just drive it straight….

The comparison was then conducted without knowing which input is which, sitting in front of the headphone amp and toggling back and forth. When necessary, I would go back and re-listen. Once I found which one sounded worse, I would then repeat the exercise by randomizing the inputs and seeing if I could still identify which one was worse. My success rate was 100% in the second test (i.e. could always verify that the first result was not by chance). This testing was repeated a number of times comparing the different sources against each other and the ML.

I did not level match anything. However, once I found one source was worse than the other, I would then turn up the volume to counter any effect there. Indeed, doing so would close the gap some but it never changed the outcome. Note that the elevated level clearly made that source sound louder than the other. So the advantage was put on the losing side.

The results above were later objectively shown to be backed by some science in Stereophile magazine. In reviews of said players and Mark Levinson, it was shown that the former would only resolve to 14 or 15 bits of audio samples. Turning off the front panel pushed some up to 16 bits or so. The ML on the other hand, was tested to have equiv. of 19.5 bits. This is contrast to all the DACs being rated at "24 bits."

Now this testing is a few years old (probably circa 2001 to 2002). Maybe DACs have improved so much that the $20 part in the player is just as good as my then $8000 Mark Levinson DAC (which was hand tuned). If so, then I like to know who has tested the new ones and details of their methodology.

There you have it. Was it worth the wait?

Quote
Originally Posted by Chu Gai
1. Am I to understand that you placed a CD into either the SACD or DVD-A and then ran an analog connection to the headphone amp followed by a digital connection from either the SACD or DVD-A to the Mark Levinson No360S followed by another analog connection to the headphone amp?

Correct. That allowed me to have near zero delay in comparisons, something that is sadly missing from HDMI world today, making any DBL test of this kind invalid in my opinion.

Quote
2. You indicate that your test probes were music, yes? Does that mean you ran original CD's or were those mp3's? If the latter what details can you provide on the means of compression.

What kind of question is that for heaven's sake. Of course these were music titles. And no, they were not MP3s. I gave the example of music compression as to have people understand that you have to use the right material for the problem you want to investigate. As with audio compression, there is material that is more revealing of DAC issues than others. And the reason for that, just like audio compression, is firmly planted in science and objective evaluation.

Quote
3. Was the Mark Levinson No360S modified and if so were you able to test a unit that was unmodified?

Modified was the wrong term to use here. No360 came in two flavors. One that came from the assembly line and the other version (the "S" I think) that was hand calibrated at ML for better performance. I seem to recall this option cost me a few thousand dollars more.

Quote
4. If it was only you doing the connections and then placing a blanket or whatever over everything, how was it that it was blind?

Pretty simple. I took advantage of old age and what it does to memory . Seriously, I would grab the RCA cables and plug them into the back without paying attention to which input was which. After the test, I would then trace the cable to the source.

In the other thread, I made a point about being honest with yourself in these tests. And that is what is going on here. All of the equipment was free to me so I had no interest to defend one or the other. I would have been just as happy to see these $1000+ sources outperform the ML so that I didn't have to use that box for my testing all the time to rule out equipment differences.

Look at it this way. If I didn't care about a fair outcome, I wouldn't even bother to go through the lengths I did to test the equipment. Raise your hand if you have two copies of multiple identical CDs, DVD-As, and SACDs. Seems like my hand is the only one up! I even have the same title in SACD and DVD-A (from Chesky).

Now let me ask you this. Have you participated in double blind tests? If so, what tests were they?

This is an old topic for you Amir.

cheers,

AJ

Notice