Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Audibility of "typical" Digital Filters in a Hi-Fi Playback  (Read 367784 times) previous topic - next topic
0 Members and 13 Guests are viewing this topic.

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

Reply #75
It's a bit like watching The View. I don't know what the UK equivalent would be.

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

Reply #76
Quote
You cannot use a single digit SPL number for your room noise.


I didn't. I gave the room noise number in terms of bits. 


What is this then Arny:
Quote from:  link=msg=880335 date=0

So what is known about the monitoring system and room used for playback? I'll bet it wasn't your typical listening room with a 40 dB SPL noise floor...

The 40 db number without any kind of spectrum analysis is psychoacoustically blind.  Isn't it? 


It's a different post, in a different context.  In the immediate context of the exact post that you claimed in your post to be responding to Amir, the noise level SPL was given in bits.  This is getting tedious.

Furthermore, it is common practice to give background noise level SPLs as a single dB number:

http://www.sengpielaudio.com/TableOfSoundPressureLevels.htm

(just one of very many on the web)




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

Reply #77
I get the feeling that "Arnold B. Krueger", "ajinfla", and "amirm" know each other - maybe in the real world, and certainly in cyberspace.

It is off topic, but if you must know, Amir and I grew up together and Arnie is our godfather.
Btw David, user profiles contain info towards who/what, as does mine.


I stand behind my HA user profile.

I just checked Amir's and its blank.

Amirm is Amir Majidimehr who came up through the worlds of Unix IT and Video editing. If memory serves he has a EE from Florida Atlantic University. He was MS's VP of Windows Digital Media whose relationship with MS ended about the same time they cancelled one of his major projects, HD DVD.

AJ, Amir and I previously spent the most time together at the AVS forums.

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

Reply #78
It's a bit like watching The View. I don't know what the UK equivalent would be.
Loose Women. I've never seen it. It's daytime TV. I assume you're being insulting

Cheers,
David.

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

Reply #79
More like this.

We hardly all agree on anything (speaker/room/treatments, passive biamping, digital disorders, etc). The only relevant thing you have to know David, is that I'm always right.

I guess this BS test shows that's it's possible to doctor up any results you desire, be it cables, amps, caps or digital.
Says nothing about the transparency of 16/44 playback itself.
Which isn't really a concern of rational people.

cheers,

AJ

Loudspeaker manufacturer

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

Reply #80
It's a bit like watching The View. I don't know what the UK equivalent would be.
Loose Women. I've never seen it. It's daytime TV. I assume you're being insulting
I vote "Big Bang Theory": a bunch of clever guys with a talent for immature social behavior

 

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

Reply #81
How many people here have actually bought the paper?

Here's a rundown of some of the relevant things I don't think have been mentioned  yet (or if they have they've largely been glossed over).

1. The aggregate result for correctly identifying a sample, across all 160 trials, is 56.25% correct.
2. When they low-passed the signal *without* quantizing it to 16-bits, they got a statistically insignificant result. Quelle surprise. Needless to say, this did not stop them declaring that the digital filters they used were clearly corrupting the sound. I shall refrain from speculating as to why they would possibly want to spread FUD about digital filters.
3. The difference between filters (when they are talking about 16-bit samples the listeners could identify) is right on the edge of statistical significance, and seems to show exactly the opposite of what they expected - their listeners seemed to find low-passing *closer* to the audioband very, very slightly less audible. They speculate that this is because the filter that cuts higher is longer, and will therefore ring more and smear things temporally.
4. They seem to think that the sampling rate imposes a time-domain resolution limitation akin to the time between two adjacent samples, which is, unless I'm very confused, a giant pile of crap.
5. Did I mention the occasionally speculatory tone?

Seriously, all this paper seems to show is that if you don't dither at all, or dither improperly, a group of carefully trained listeners will, on some sample material, be able to distinguish the 16-bit content from the original content, albeit with a fair-to-great amount of difficulty (they did much better worse with rectangular dither than with no dither). Some small differences with distinctly diminished sample sizes are then used to come to some very strange conclusions. There's all sorts of guff about temporal smearing, and a ramble through some decidedly dubious literature (yes, Ooashi et al make an appearance) to portray the 'prima facies' case for hi-res as a reasonably solid one.

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

Reply #82
all this paper seems to show is that if you don't dither at all, or dither improperly, a group of carefully trained listeners will, on some sample material, be able to distinguish the 16-bit content from the original content

Right, and most of us were already aware this was possible.

temporal smearing

I imagine they gotta keep selling the case for magical apodizing filters.

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

Reply #83
Amirm is Amir Majidimehr who came up through the worlds of Unix IT and Video editing. If memory serves he has a EE from Florida Atlantic University. He was MS's VP of Windows Digital Media whose relationship with MS ended about the same time they cancelled one of his major projects, HD DVD.  AJ, Amir and I previously spent the most time together at the AVS forums.
  Arny, you forgot to mention an important career move of his that 2Bdecided might consider relevant, namely that Amir founded an audio retail store which sells, among other things, so-called "high end"  amps, processors, and I would assume the cables to hook them up. His AVS forum posts mention he is "retired", perhaps in an attempt to mitigate any potential claims of him having a pecuniary interest in promoting his high end industry and/or its ideologies, however the webpage of his company which his posts' signature directly links to (over 18,000 posts, by the way) makes no mention of him being "retired" from that particular company:  http://www.madronadigital.com/about/About.html

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

Reply #84
I have followed the execution and discussion of several comparative tests recently, starting with Arnie's keys. All of the tests proved to have samples that were defective. I believe this is the issue that needs to be discussed, not the results of the tests. GI = GO.



aka 'train wrecks'.   


It appears the wreckage is still far from being cleared.  Some even seem to just spread the wreckage around, rather than clean it up. 



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

Reply #85
Right.

Ok, we've made introductions and are aware of the history.  Future posts will be on the topic covered in the first post.

@armirm:
I'm sorry you missed this latest round, but the window to share your background, etc. is over.  Whether or not you feel you've been given a fair shake is not the forum's problem.

@all:
Segues into amps, speakers, cables, personalities, backgrounds, who said what when and other proxy discussions (including the AVS debacle!) will be binned.  Continued discussion will result in a curtailment of posting privileges.

If you wish to argue other topics, start a new discussion in the appropriate forum if one doesn't already exist.  The same rules will apply there as well.



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

Reply #88
2. When they low-passed the signal *without* quantizing it to 16-bits, they got a statistically insignificant result.
Did they specify the low-pass filter(s) for that test ?

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

Reply #89
How many people here have actually bought the paper?

I am an AES member and so I have the paper.

Quote
Here's a rundown of some of the relevant things I don't think have been mentioned  yet (or if they have they've largely been glossed over).

I am pretty sure there are more relevant things than what you listed .

Quote
Seriously, all this paper seems to show is that if you don't dither at all, or dither improperly, a group of carefully trained listeners will, on some sample material, be able to distinguish the 16-bit content from the original content, albeit with a fair-to-great amount of difficulty (they did much better worse with rectangular dither than with no dither).

You just gave away the farm with that summary.  If we are sensitive to such things as dither types then all the talk about any and all small distortions not being audible is proven to be wrong.  These are small differences. Right?  If we can hear them then we can hear a lot of other artifacts we shove under the rug.

As to "carefully trained listeners," that is a sign of merit for any such tests.  Using untrained listeners goes against the best industry practices as outlined in ITU BS-1116.  The training here was very limited to this one test alone.  The listening group were not trained experts.  They were allowed to listen to the music and hear a controlled FIR filter test.

Quote
Some small differences with distinctly diminished sample sizes are then used to come to some very strange conclusions. There's all sorts of guff about temporal smearing, and a ramble through some decidedly dubious literature (yes, Ooashi et al make an appearance) to portray the 'prima facies' case for hi-res as a reasonably solid one.

Seems like you read it in an emotional state of mind given this description.  I would love to see your read of Meyer and Moran so that I get calibrated on how you judge such things.  Is there a link for that?
Amir
Retired Technology Insider
Founder, AudioScienceReview.com

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

Reply #90
@armirm:
I'm sorry you missed this latest round, but the window to share your background, etc. is over.  Whether or not you feel you've been given a fair shake is not the forum's problem.

Hmmm.  I am bombarded with posts addressing me specifically in multiple threads.  I am trying to do my best to answer them.  I see David's request now but didn't realize there was a time limit on answering it.  Closing the opportunity for me to answer them is forum's problem.  But it is your forum, you get to set the rules and my job is to follow them.  And so shall I .
Amir
Retired Technology Insider
Founder, AudioScienceReview.com

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

Reply #91
all this paper seems to show is that if you don't dither at all, or dither improperly, a group of carefully trained listeners will, on some sample material, be able to distinguish the 16-bit content from the original content

Right, and most of us were already aware this was possible.

I haven't seen this awareness in other forums.  Is there a link here where folks have said that in double blind tests, people can tell the difference between how 24 bit files are or are not dithered to 16 bits?

Quote
temporal smearing

I imagine they gotta keep selling the case for magical apodizing filters.

And what is our theory of how listeners could tell filters apart in this blind test?
Amir
Retired Technology Insider
Founder, AudioScienceReview.com

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

Reply #92
If we are sensitive to such things as dither types then all Strawman the argument to talk about any and all small distortions not being audible is proven to be wrong
If we Red Herring the argument to can hear them then we can hear a lot of other artifacts we shove under the rug

I would love Red Herring the argument to to see your read of Meyer and Moran so that.....

FIFY.
Loudspeaker manufacturer

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

Reply #93
Back on topic: for those who missed it, a test of filter audibility has already been attempted right here on HA...
http://www.hydrogenaud.io/forums/index.php?showtopic=68524

Any new takers?

Cheers,
David.

Excellent!  Two positive outcomes.  I stand corrected.  You all have tested filters and know the differences can be heard in ABX blind tests.  So why all the angst over Stuart's formal test of the same?

And why is it that when the same posters come to AVS forum they don't acknowledge such data?  I may have missed it of course but prior to this round of testing, no one ever had posted outputs of foobar with positive detection of small differences such as we are talking about.

But kudos to you and HA forum for creating the data and opportunity. 

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%)



As an aside, I really enjoy reading your posts.  Clearly you know your stuff and stay out of the emotional components.
Amir
Retired Technology Insider
Founder, AudioScienceReview.com

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

Reply #94
And here is the same using the new Foobar 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
Amir
Retired Technology Insider
Founder, AudioScienceReview.com

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

Reply #95
Your computation there is psychoacoustically blind Arny as we have discussed.  You cannot use a single digit SPL number for your room noise.  You must look at its spectrum and see how much noise you have relative to threshold of hearing as I show in my article
I agree with the need to look at the spectrum, but disagree with your conclusions.

http://www.hydrogenaud.io/forums/index.php...st&p=871182
http://www.hydrogenaud.io/forums/index.php...st&p=871270

It's really hard to find source material that has audible noise when you use 16-bit noise shaped dither. The current (Bob Stuart) test apparently used no noise shaping, and the wrong dither PDF.

Cheers,
David.

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

Reply #96
And verification of the signature:

Signature matches; the log appears to be valid.

For some reason cutting and pasting from the forum doesn't work.  I had to feed it the text file that was stored.  Maybe it is me adding formatting to it.
Amir
Retired Technology Insider
Founder, AudioScienceReview.com

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

Reply #97
08:44:21 : 01/04  93.8%  <--- Difference found.
What does it sound like?

And verification of the signature:

Signature matches; the log appears to be valid.
That's fair enough, but anyone who suspects cheating is unlikely to think that edited ABX logs are the way someone would do it.

Cheers,
David.

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

Reply #98
Your computation there is psychoacoustically blind Arny as we have discussed.  You cannot use a single digit SPL number for your room noise.  You must look at its spectrum and see how much noise you have relative to threshold of hearing as I show in my article
I agree with the need to look at the spectrum, but disagree with your conclusions.

http://www.hydrogenaud.io/forums/index.php...st&p=871182
http://www.hydrogenaud.io/forums/index.php...st&p=871270

It's really hard to find source material that has audible noise when you use 16-bit noise shaped dither. The current (Bob Stuart) test apparently used no noise shaping, and the wrong dither PDF.

Cheers,
David.

Sorry David. I read your link and it almosts reads word for word like my post and article.  What is it that I am missing?

Meanwhile, the subject is even more complex than my brief write-up.  Noise that comes from the source is played through point sources of our speakers.  Listening tests show that we can hear point noise over broad noise that is resident in the room.  This is no doubt due to evolution of human beings.  A dangerous noise coming from a specific point trumps broad noises of the environment.  There would be fewer of us if that had not been true and more of our ancestors had gotten eaten .

As to your comment regarding content, it is fine for such existence to be rare.  We can afford to preserve all that can be for ALL content for no cost today.  I don't want to analyze each piece of music I get to determine whether something was stepped on.  We can get a format that is transparent (> great than CD specs) and be done with all of these technical arguments.  One only has to understand this topic if it is to justify lowering the resolution of the content.  We are in no need of such lowering technologically and from business point of view.
Amir
Retired Technology Insider
Founder, AudioScienceReview.com

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

Reply #99
Meanwhile, the subject is even more complex than my brief write-up.  Noise that comes from the source is played through point sources of our speakers.


Raising the question of how does one distinguish a particular noise in the source from any of the many sources of noise in the source?  They all come to our ears from the same speakers...

Also, modern audio systems are surround systems, which reproduce sound from up to 11 or more separate point sources spread throughout the room.  Their purpose is to recreate a sonic image that existed in the performance space here the recording was made, which means that they must blend together into a continuous sound feel that duplicates what one would hear in a room.  How does one reliably distinguish that continuous sound field from the recording from the one created by various noise sources in the room?

I suspect that the well-known asymmetries in the recent Meridian tests may be tacit admission that without adding them, the listening tests would have been expected by the paper's perpetrators to produce null reslts.