IPB

Welcome Guest ( Log In | Register )

Multiformat listening test @ ~64kbps: Results, Results and post-test discussion
IgorC
post Apr 12 2011, 00:40
Post #1





Group: Members
Posts: 1572
Joined: 3-January 05
From: ARG/RUS
Member No.: 18803



The test is finished, results are available here:

http://listening-tests.hydrogenaudio.org/igorc/results.html

Summary: CELT/Opus won, Apple HE-AAC is better than Nero HE-AAC, and Vorbis has caught up with Nero HE-AAC.
Go to the top of the page
+Quote Post
 
Start new topic
Replies
C.R.Helmrich
post Apr 12 2011, 20:11
Post #2





Group: Developer
Posts: 686
Joined: 6-December 08
From: Erlangen Germany
Member No.: 64012



Thanks for organizing the tests, guys! Sorry for being picky, but I'm not convinced about the analysis. To ease my mind, it would be great if you could comment on the following.

  • Please provide the number of valid results (i.e. listeners) per sample (excluding "27", see below).
  • How did you compute the overall average score of a codec and its confidence intervals? Taking the mean of all listeners' results? That would mean a sample with more listeners (i.e. probably sample01) has a greater influence than the last few samples (which still needed listeners shortly before the end of the test). This is probably not a good approach; weighting each sample equally in the overall score seems to be the way to go for me (but it probably doesn't make a difference here, but still...).
  • Nothing personal, but if a listener like "27" consistently scores in opposite direction as the average (as shown by Igor), a thorough post-screening analysis (like Spearman rank correlation < some value) would - and has to - exclude such results.


Edit: Christoph, why are the samples you uploaded at 96 kHz? Did you do the test that way?

Chris

This post has been edited by C.R.Helmrich: Apr 12 2011, 20:17


--------------------
If I don't reply to your reply, it means I agree with you.
Go to the top of the page
+Quote Post
NullC
post Apr 12 2011, 21:47
Post #3





Group: Developer
Posts: 200
Joined: 8-July 03
Member No.: 7653



QUOTE (C.R.Helmrich @ Apr 12 2011, 11:11) *
Sorry for being picky, but I'm not convinced about the analysis.


The paired statistical tests are pretty incontrovertible. I've since run the same analysis with a number of different balancing and post filtering rules and every time it's come out to be the same way.

If it's any consolation, Opus considerably bombs the couple of cases that it does poorly (though its sample by sample variance is still not as large as the other codecs, it has stronger outliers). This is undoubtedly due to a mixture of encoder immaturity, lack of taking advantage of VBR, and just one of the annoying tradeoffs that come from creating a low latency codec. (The mode opus was used in here has a total of 22.5ms of latency, including the overlap but ignoring any serialization delay related to VBR).

I've noticed that there seems to be some misunderstanding promoted around here related to confidence intervals. Even ignoring the issues with non-pairwise comparisons, assumptions of normality, etc. there seems to be a mis-aprehension that the confidence intervals must not overlap at all for the result to be deemed significant to whatever P-value was used to draw the bars. This is clearly incorrect.

For example, consider 5% error bars on the mean of codec A and 5% bars on the mean codec B and the lower bar of A is the same as the upper bar of B. Is there a 1/20 (p=0.05) chance that the difference in means arose from noise? _NO_ If we assume that the errors are independent the chance of that is more like 1/400 (0.05^2). Of course, the errors are not completely independently distributed— but this fact also invalidates the assumptions used to set the errors bars in the first place. Another approach would be to compare the mean of one value with the error-bars on the mean of the other and vice versa, this isn't ideal either but it does avoid squaring the P-value used.

Blocked pair-wise parametric tests are much better for this reason, and others but they don't result in pretty graphs.

This post has been edited by NullC: Apr 12 2011, 21:48
Go to the top of the page
+Quote Post

Posts in this topic
- IgorC   Multiformat listening test @ ~64kbps: Results   Apr 12 2011, 00:40
- - Garf   If someone can assist with a bitrate table or per-...   Apr 12 2011, 01:02
- - Garf   Oh, and given that Opus is open sourced, if one of...   Apr 12 2011, 01:06
- - AllanP   I just wonder one thing, when the Vorbis encoder w...   Apr 12 2011, 01:14
|- - Garf   QUOTE (AllanP @ Apr 12 2011, 02:14) I jus...   Apr 12 2011, 01:15
|- - AllanP   QUOTE (Garf @ Apr 12 2011, 02:15) You can...   Apr 12 2011, 01:22
- - romor   Congratulation to CELT/Opus! I wanted to com...   Apr 12 2011, 03:08
- - IgorC   I think the results of lessthanjoey and AlexB are ...   Apr 12 2011, 03:50
- - googlebot   I'm stunned by the CELT/Opus results! I wo...   Apr 12 2011, 08:06
|- - NullC   QUOTE (googlebot @ Apr 11 2011, 23:06) I...   Apr 14 2011, 05:00
|- - saratoga   QUOTE (NullC @ Apr 14 2011, 00:00) We als...   Apr 14 2011, 06:29
||- - NullC   QUOTE (saratoga @ Apr 13 2011, 22:29) Is ...   Apr 14 2011, 08:30
|- - Garf   QUOTE (NullC @ Apr 14 2011, 06:00) Low-la...   Apr 14 2011, 12:19
|- - jmvalin   QUOTE (Garf @ Apr 14 2011, 07:19) Is the ...   Apr 14 2011, 14:04
|- - NullC   QUOTE (Garf @ Apr 14 2011, 03:19) QUOTE (...   Apr 14 2011, 17:47
|- - C.R.Helmrich   QUOTE (NullC @ Apr 14 2011, 18:47) The sw...   Apr 14 2011, 23:39
|- - jmvalin   QUOTE (C.R.Helmrich @ Apr 14 2011, 18:39)...   Apr 15 2011, 05:49
- - Alex B   Thanks guys! Interesting results. One note t...   Apr 12 2011, 12:59
|- - Garf   QUOTE (Alex B @ Apr 12 2011, 13:59) I got...   Apr 12 2011, 13:53
|- - NullC   QUOTE For processing the result .txt files with ch...   Apr 12 2011, 14:48
|- - Garf   QUOTE (NullC @ Apr 12 2011, 15:48) Sounds...   Apr 12 2011, 14:59
||- - Alex B   QUOTE (Garf @ Apr 12 2011, 16:59) But the...   Apr 12 2011, 15:09
|- - Alex B   QUOTE (NullC @ Apr 12 2011, 16:48) Sounds...   Apr 12 2011, 15:02
|- - NullC   QUOTE (Alex B @ Apr 12 2011, 07:02) QUOTE...   Apr 12 2011, 15:17
|- - motion_blur   QUOTE (Alex B @ Apr 12 2011, 16:02) QUOTE...   Apr 12 2011, 16:15
|- - NullC   QUOTE (motion_blur @ Apr 12 2011, 08:15) ...   Apr 12 2011, 17:54
|- - motion_blur   QUOTE (NullC @ Apr 12 2011, 18:54) QUOTE ...   Apr 12 2011, 19:42
- - Alex B   For comparison I uploaded a rar package of my ...   Apr 12 2011, 14:14
|- - Garf   QUOTE (Alex B @ Apr 12 2011, 15:14) For c...   Apr 12 2011, 14:49
- - Alex B   QUOTE (Garf @ Apr 12 2011, 16:59) But the...   Apr 12 2011, 15:35
|- - Garf   QUOTE (Alex B @ Apr 12 2011, 16:35) QUOTE...   Apr 12 2011, 15:42
- - Alex B   Regarding the bitrate table, I guess that CELT/Op...   Apr 12 2011, 16:14
|- - NullC   QUOTE (Alex B @ Apr 12 2011, 08:14) Regar...   Apr 12 2011, 18:10
- - IgorC   Yes, I was too strict. Sorry about it. Some of th...   Apr 12 2011, 18:13
|- - motion_blur   QUOTE (IgorC @ Apr 12 2011, 19:13) Yes, I...   Apr 12 2011, 20:09
|- - NullC   QUOTE (motion_blur @ Apr 12 2011, 12:09) ...   Apr 13 2011, 00:59
|- - motion_blur   QUOTE (NullC @ Apr 13 2011, 01:59) QUOTE ...   Apr 13 2011, 10:06
- - markanini   I figured ratings would vary between testers depen...   Apr 12 2011, 18:52
|- - NullC   QUOTE (markanini @ Apr 12 2011, 09:52) I ...   Apr 12 2011, 20:57
- - lessthanjoey   I've done some more testing with headphones af...   Apr 12 2011, 19:51
- - C.R.Helmrich   Thanks for organizing the tests, guys! Sorry f...   Apr 12 2011, 20:11
|- - motion_blur   QUOTE (C.R.Helmrich @ Apr 12 2011, 21:11)...   Apr 12 2011, 20:41
|- - Garf   QUOTE Please provide the number of valid results (...   Apr 12 2011, 21:44
||- - C.R.Helmrich   Sorry, Christoph, can't reproduce it. What you...   Apr 12 2011, 22:12
||- - Garf   QUOTE (C.R.Helmrich @ Apr 12 2011, 23:12)...   Apr 12 2011, 23:37
||- - motion_blur   QUOTE (C.R.Helmrich @ Apr 12 2011, 23:12)...   Apr 13 2011, 00:41
|- - NullC   QUOTE (C.R.Helmrich @ Apr 12 2011, 11:11)...   Apr 12 2011, 21:47
- - IgorC   motion_blur, You can download the results of all...   Apr 12 2011, 20:21
|- - _mē_   Some presentation suggestions: 1. Codec versions a...   Apr 12 2011, 20:43
|- - IgorC   QUOTE (_mē_ @ Apr 12 2011, 16:43) Some pr...   Apr 12 2011, 20:46
|- - Alex B   QUOTE (_mē_ @ Apr 12 2011, 22:43) 2. Link...   Apr 12 2011, 21:06
||- - Alex B   QUOTE (Alex B @ Apr 12 2011, 23:06) QUOTE...   Apr 13 2011, 11:38
||- - NullC   QUOTE (Alex B @ Apr 13 2011, 02:38) The b...   Apr 13 2011, 12:41
||- - Garf   QUOTE (NullC @ Apr 13 2011, 13:41) QUOTE ...   Apr 13 2011, 12:54
||- - Alex B   QUOTE (NullC @ Apr 13 2011, 14:41) Any id...   Apr 13 2011, 14:46
||- - NullC   QUOTE (Alex B @ Apr 13 2011, 05:46) QUOTE...   Apr 13 2011, 22:48
|- - Garf   QUOTE (_mē_ @ Apr 12 2011, 21:43) Some pr...   Apr 12 2011, 22:24
- - Alex B   Here is the raw data for a bitrate table. The bitr...   Apr 12 2011, 20:30
- - IgorC   Thank you for your help, AlexB. If you can do the ...   Apr 12 2011, 20:44
- - C.R.Helmrich   Christoph, do you mean the slightly washed out bas...   Apr 12 2011, 20:52
|- - motion_blur   QUOTE (C.R.Helmrich @ Apr 12 2011, 21:52)...   Apr 12 2011, 21:11
|- - C.R.Helmrich   QUOTE (motion_blur @ Apr 12 2011, 22:11) ...   Apr 12 2011, 21:38
|- - motion_blur   QUOTE (C.R.Helmrich @ Apr 12 2011, 22:38)...   Apr 12 2011, 21:56
- - IgorC   I've checked. The decoder on Christoph's s...   Apr 12 2011, 20:54
- - Alex B   Here's the bitrate table: In Excel format: ...   Apr 12 2011, 22:20
|- - saintdev   QUOTE (Alex B @ Apr 12 2011, 14:20) Here...   Apr 12 2011, 23:21
- - NullC   QUOTE (IgorC @ Apr 11 2011, 16:40) The te...   Apr 13 2011, 04:42
|- - Garf   QUOTE (NullC @ Apr 13 2011, 05:42) Hey al...   Apr 13 2011, 08:27
||- - IgorC   QUOTE (Garf @ Apr 13 2011, 04:27) One thi...   Apr 14 2011, 05:57
|- - C.R.Helmrich   Thanks, Garf, for the plot! And thanks, Christ...   Apr 13 2011, 22:46
|- - Garf   QUOTE (C.R.Helmrich @ Apr 13 2011, 23:46)...   Apr 13 2011, 22:57
||- - C.R.Helmrich   QUOTE (Garf @ Apr 13 2011, 23:57) From th...   Apr 13 2011, 23:06
|- - jmvalin   QUOTE (C.R.Helmrich @ Apr 13 2011, 17:46)...   Apr 14 2011, 00:58
|- - Garf   QUOTE (jmvalin @ Apr 14 2011, 01:58) I do...   Apr 14 2011, 09:13
|- - jmvalin   QUOTE (Garf @ Apr 14 2011, 04:13) You are...   Apr 14 2011, 11:41
- - Garf   The result page is now updated with per-sample gra...   Apr 13 2011, 12:41
- - mixminus1   Thanks much to all for their work in both setting ...   Apr 13 2011, 14:55
|- - Garf   QUOTE (mixminus1 @ Apr 13 2011, 15:55) Th...   Apr 13 2011, 15:04
- - mixminus1   :facepalm: Good God... Thanks, Garf, I was scour...   Apr 13 2011, 15:15
- - romor   @Garf: can you please reupload results you posted ...   Apr 13 2011, 16:35
|- - Garf   QUOTE (romor @ Apr 13 2011, 17:35) @Garf:...   Apr 13 2011, 18:59
- - IgorC   AlexB, thank you for bitrate verification. I real...   Apr 13 2011, 17:12
- - pdq   Whether or not classical should be considered to b...   Apr 13 2011, 17:51
|- - IgorC   QUOTE (pdq @ Apr 13 2011, 13:51) Whether ...   Apr 13 2011, 18:01
|- - pdq   QUOTE (IgorC @ Apr 13 2011, 13:01) QUOTE ...   Apr 13 2011, 18:53
- - romor   file: http://people.xiph.org/~greg/opus/ha2011/2.....   Apr 13 2011, 19:28
|- - NullC   QUOTE (romor @ Apr 13 2011, 10:28) file: ...   Apr 13 2011, 19:53
- - IgorC   Bitrate verification on my set of albums: http:/...   Apr 13 2011, 21:45
|- - NullC   QUOTE (IgorC @ Apr 13 2011, 13:45) Bitrat...   Apr 21 2011, 19:15
|- - jmvalin   QUOTE (NullC @ Apr 21 2011, 14:15) QUOTE ...   Apr 21 2011, 19:53
- - Garf   QUOTE Has anyone ever seriously blind-tested e.g. ...   Apr 14 2011, 00:10
|- - C.R.Helmrich   QUOTE (Garf @ Apr 14 2011, 01:10) I'm...   Apr 14 2011, 10:46
- - .alexander.   The second graph seems to be consistent with ...   Apr 15 2011, 12:40
|- - jmvalin   QUOTE (.alexander. @ Apr 15 2011, 07:40) ...   Apr 15 2011, 15:17
- - Xanikseo   QUOTE (IgorC @ Apr 13 2011, 21:45) Bitrat...   Apr 20 2011, 16:34
|- - Garf   QUOTE (Xanikseo @ Apr 20 2011, 17:34) Igo...   Apr 20 2011, 18:02
|- - IgorC   QUOTE (Xanikseo @ Apr 20 2011, 12:34) Igo...   Apr 20 2011, 18:19
|- - Zarggg   QUOTE (Xanikseo @ Apr 20 2011, 11:34) EDI...   Apr 20 2011, 19:09
- - Xanikseo   QUOTE (Zarggg @ Apr 20 2011, 19:09) QUOTE...   Apr 20 2011, 21:20
- - IgorC   NullC, h*tp://www.mediafire.com/?s7i9usu2qr27pcg ...   Apr 21 2011, 23:52
2 Pages V   1 2 >


Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 15th September 2014 - 11:26