IPB

Welcome Guest ( Log In | Register )

2 Pages V   1 2 >  
Reply to this topicStart new topic
Comparing two Nero AAC settings
muaddib
post Aug 10 2009, 12:34
Post #1





Group: Developer
Posts: 398
Joined: 14-October 01
Member No.: 289



Anybody willing to compare two versions of Nero AAC Encoder?
In the attachment are two samples (original and 2 encoded versions for each). These are famous bibilolo and emese samples.
Both are encoded using abr 224 kbps (there is slight bitrate deviation). Please ignore this deviation.
And please don't look at spectrum plots. Please, no posting of spectrum plots in this thread or any discussion about spectrum!
We are interested if any of you can hear any difference and which encoder version produces a sample that sounds closer to the original.

This post has been edited by muaddib: Aug 10 2009, 12:34
Attached File(s)
Attached File  56_12_vs_56_13.zip ( 4.37MB ) Number of downloads: 531
 
Go to the top of the page
+Quote Post
vpa
post Aug 10 2009, 13:12
Post #2





Group: Members
Posts: 99
Joined: 13-April 08
Member No.: 52748



QUOTE (muaddib @ Aug 10 2009, 13:34) *
Anybody willing to compare two versions of Nero AAC Encoder?
In the attachment are two samples (original and 2 encoded versions for each). These are famous bibilolo and emese samples.
Both are encoded using abr 224 kbps (there is slight bitrate deviation). Please ignore this deviation.
And please don't look at spectrum plots. Please, no posting of spectrum plots in this thread or any discussion about spectrum!
We are interested if any of you can hear any difference and which encoder version produces a sample that sounds closer to the original.

Not an ABX Result, but to me the second one seems to extend a lttle more to frequency extremes (deeper bass, more realistic trebble), but there is also a little bit of ringing sound?


--------------------
WavPack 4.50.1 -hhx6 | LAME 3.98.2 -V 0
Go to the top of the page
+Quote Post
rpp3po
post Aug 10 2009, 13:13
Post #3





Group: Developer
Posts: 1126
Joined: 11-February 03
From: Germany
Member No.: 4961



I have just tried the emese samples. In my opinion both encodes fail pretty badly. There is very loud sibilance rapidly altering with the track's rhythmic percussion. If I had not known the original I would have believed that it was a purposely applied special effect or additional electronic instrument. But compared to the original I must say this is one of the loudest artefacts I have ever heard.

I have focussed on the range 0:02-0:05. These are the results. Regarding above artifact 13 is no improvement over 12 or vice versa. Because this was so dominant I have not compared other "layers" of the track.
CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.8
2009/08/10 13:55:13

File A: Z:\rpp3po\Downloads\56_12_vs_56_13\emese.flac
File B: Z:\rpp3po\Downloads\56_12_vs_56_13\emese_56_12.mp4

13:55:13 : Test started.
13:56:52 : 01/01 50.0%
13:56:57 : 02/02 25.0%
13:57:06 : 03/03 12.5%
13:57:10 : 04/04 6.3%
13:57:19 : 05/05 3.1%
13:57:26 : 06/06 1.6%
13:57:42 : 07/07 0.8%
13:57:48 : 08/08 0.4%
13:57:54 : 09/09 0.2%
13:58:09 : 10/10 0.1%
13:58:11 : Test finished.

----------
Total: 10/10 (0.1%)

CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.8
2009/08/10 13:51:10

File A: Z:\rpp3po\Downloads\56_12_vs_56_13\emese.flac
File B: Z:\rpp3po\Downloads\56_12_vs_56_13\emese_56_13.mp4

13:51:10 : Test started.
13:53:28 : 01/01 50.0%
13:53:36 : 02/02 25.0%
13:53:42 : 03/03 12.5%
13:53:51 : 04/04 6.3%
13:54:02 : 05/05 3.1%
13:54:09 : 06/06 1.6%
13:54:16 : 07/07 0.8%
13:54:27 : 08/08 0.4%
13:54:36 : 09/09 0.2%
13:54:45 : 10/10 0.1%
13:54:48 : Test finished.

----------
Total: 10/10 (0.1%)


This post has been edited by rpp3po: Aug 10 2009, 13:15
Go to the top of the page
+Quote Post
muaddib
post Aug 10 2009, 13:21
Post #4





Group: Developer
Posts: 398
Joined: 14-October 01
Member No.: 289



These two samples were chosen because many people can hear difference from the originals.
What we are concerned in this test is whether there is any difference between these two encoded versions that one can hear.
Go to the top of the page
+Quote Post
sauvage78
post Aug 10 2009, 13:22
Post #5





Group: Members
Posts: 677
Joined: 4-May 08
Member No.: 53282



I tried the emese samples as I already knew this one, I can ABX both versions against the original, but I cannot ABX one nero AAC versions against the other, for me both are not transparent & same.

Below are the logs against the original which shows that I can clearly ABX the artefact (noise in the back), but there is no log for 12 Vs. 13 as I gave up after ... say 30 listening trials because I couldn't decide without guessing. Edit: I only used second 0 to 3 for my test.

CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.8
2009/08/10 14:05:46

File A: C:\Documents and Settings\Sauvage\Bureau\56_12_vs_56_13\emese.flac
File B: C:\Documents and Settings\Sauvage\Bureau\56_12_vs_56_13\emese_56_12.mp4

14:05:46 : Test started.
14:06:08 : 01/01 50.0%
14:06:23 : 02/02 25.0%
14:06:49 : 03/03 12.5%
14:07:03 : 04/04 6.3%
14:07:14 : 05/05 3.1%
14:07:32 : 06/06 1.6%
14:07:43 : 07/07 0.8%
14:07:54 : 08/08 0.4%
14:07:58 : Test finished.

----------
Total: 8/8 (0.4%)


CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.8
2009/08/10 14:09:28

File A: C:\Documents and Settings\Sauvage\Bureau\56_12_vs_56_13\emese.flac
File B: C:\Documents and Settings\Sauvage\Bureau\56_12_vs_56_13\emese_56_13.mp4

14:09:28 : Test started.
14:09:50 : 01/01 50.0%
14:10:02 : 02/02 25.0%
14:10:12 : 03/03 12.5%
14:10:26 : 04/04 6.3%
14:10:36 : 05/05 3.1%
14:10:50 : 06/06 1.6%
14:11:04 : 07/07 0.8%
14:11:30 : 08/08 0.4%
14:11:33 : Test finished.

----------
Total: 8/8 (0.4%)


Moderation: Codeboxed.

This post has been edited by JensRex: Aug 10 2009, 14:08


--------------------
CDImage+CUE
Secure [Low/C2/AR(2)]
Flac -4
Go to the top of the page
+Quote Post
rpp3po
post Aug 10 2009, 13:25
Post #6





Group: Developer
Posts: 1126
Joined: 11-February 03
From: Germany
Member No.: 4961



I also can't ABX both AAC encodes against each other, yet. Quicktime 7.6.2 also fails pretty badly at true VBR Q127. The sample is pretty though, is there anything able to handle it, yet?

This post has been edited by rpp3po: Aug 10 2009, 14:29
Go to the top of the page
+Quote Post
muaddib
post Aug 10 2009, 13:28
Post #7





Group: Developer
Posts: 398
Joined: 14-October 01
Member No.: 289



For bibilolo, one should be able to hear some problems in background noise, when the volume is turned up.
At least they are clear to me at lower bitrates (~192kbps).
Go to the top of the page
+Quote Post
rpp3po
post Aug 10 2009, 13:49
Post #8





Group: Developer
Posts: 1126
Joined: 11-February 03
From: Germany
Member No.: 4961



I have tried bibilolo. Much harder than emese, but I can hear additional background noise when I crank up the volume high enough. I have tried very hard until my ears rang, but I can't hear any difference between both AAC tracks.

This post has been edited by rpp3po: Aug 10 2009, 13:51
Go to the top of the page
+Quote Post
sauvage78
post Aug 10 2009, 20:31
Post #9





Group: Members
Posts: 677
Joined: 4-May 08
Member No.: 53282



As I am not trained on the bibilolo sample I tried with V1.3.3.0 at lower bitrates 0.40/0.55/0.60 aka 128/192/224Kbps VBR, I could instantly ABX 128 & 192 but failed at 224Kbps. I spot a noise at 02.9 second exactly, I didn't listen any further. I cannot ABX this noise on any of the two samples you provided. Maybe there is something ABXable on the rest of the sample, I dunno.

According to me you used a bitrate that is too high, the difference between the samples is almost un-abxable if there is any.

Comparing V1.3.3.0 -q 0.60 against 12 & 13 I suspect that 12 & 13 sound better, as I was almost able to ABX a slight something with official nero, the edge of transparency is very close IMHO. But the bitrate is not the same it's 168Kbps VBR against 240Kbps ABR, so I cannot tell if the suspected improvement is due to a better encoder or if it's simply due to a higher bitrate. I suspect it's due to higher bitrate simply. Anyway I cannot back up these suspitions with ABX logs.

Overall, for me, there is no difference between 12 & 13 on the small portion of the samples I tested (The 3 first seconds for both). I don't plan to test further.

Edit:
If someone as more info on the bibilolo sample plz post them, I am searching Artist/Album/Song, it's usefull to identify the kind of music that is hard to encode. Obviously it's some kind of ambience music but if possible I would like to know exactly in case I would use this sample in a later listening test, thks.

This post has been edited by sauvage78: Aug 10 2009, 21:00


--------------------
CDImage+CUE
Secure [Low/C2/AR(2)]
Flac -4
Go to the top of the page
+Quote Post
C.R.Helmrich
post Aug 11 2009, 07:40
Post #10





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



Can't hear any difference between _12 and _13 either. At least not with headphones. Might try again with loudspeakers tonight.

Chris


--------------------
If I don't reply to your reply, it means I agree with you.
Go to the top of the page
+Quote Post
muaddib
post Aug 13 2009, 07:44
Post #11





Group: Developer
Posts: 398
Joined: 14-October 01
Member No.: 289



Thanks to all people that did the test.
If somebody else is willing to do the test, the fastest would be to just try to ABX two .mp4 files. If you don't hear any difference between them, just let us know and it will be a big help.
The original files are there to make it easier for you to find problematic parts and to compare .mp4s to the original if you can ABX two .mp4 versions.
Go to the top of the page
+Quote Post
IgorC
post Aug 13 2009, 09:03
Post #12





Group: Members
Posts: 1554
Joined: 3-January 05
From: ARG/RUS
Member No.: 18803



I've tried sample 'emeses'. Source vs Encoded (v12 and v13) was easy to ABX. Encoded v12 vs Encoded v13 wasn't easy at all.
I was able to hear difference in training ff123's ABX mode (5 of 5 chances) between both versions v12 and v13 but the difference is so subtle that when I tried in normal mode to do log file I was already desconcentrated but still was close 4/5 and second chance already worse 3/5.
Twice time I prefered version 12. Its artifacts are more soft and more pleasant to ear. While artifacts of version 13 were physically more annoying.


Training ABX mode was succeed (5 of 5) . Not logged. sad.gif

1st try (4 of 5):

CODE
ABC/HR Version 1.1 beta 2, 18 June 2004
Testname:

1R = C:\0\NERO Developres\emeses\emese_56_13.wav
2R = C:\0\NERO Developres\emeses\emese_56_12.wav

---------------------------------------
General Comments:

---------------------------------------
1R File: C:\0\NERO Developres\emeses\emese_56_13.wav
1R Rating: 4.5
1R Comment:
---------------------------------------
2R File: C:\0\NERO Developres\emeses\emese_56_12.wav
2R Rating: 4.7
2R Comment:
---------------------------------------
ABX Results:
C:\0\NERO Developres\emeses\emese_56_13.wav vs C:\0\NERO Developres\emeses\emese_56_12.wav
    4 out of 5, pval = 0.188


2nd try (3 of 5):
CODE
ABC/HR Version 1.1 beta 2, 18 June 2004
Testname:

1R = C:\0\NERO Developres\emeses\emese_56_13.wav
2R = C:\0\NERO Developres\emeses\emese_56_12.wav

---------------------------------------
General Comments:

---------------------------------------
1R File: C:\0\NERO Developres\emeses\emese_56_13.wav
1R Rating: 4.5
1R Comment:
---------------------------------------
2R File: C:\0\NERO Developres\emeses\emese_56_12.wav
2R Rating: 4.7
2R Comment:
---------------------------------------
ABX Results:
C:\0\NERO Developres\emeses\emese_56_13.wav vs C:\0\NERO Developres\emeses\emese_56_12.wav
    3 out of 5, pval = 0.500

I hope my results somehow maybe usefull but if it's not feel free to ignore them.

This post has been edited by IgorC: Aug 13 2009, 09:16
Go to the top of the page
+Quote Post
Alexxander
post Aug 13 2009, 21:20
Post #13





Group: Members
Posts: 457
Joined: 15-November 04
Member No.: 18143



I tried the emese sample, abx-ing against flac was not hard when focussing on 0:00.5-0:02.1 or 0:03.3-0:04.9. Within the first mentioned range the lossy samples sound with less detail, a bit smeared (if this is the appropriate term) while the second mentioned range was easier to ABX because the lossy samples have some kind of harmonic (higher frequency) distortion. Comparing both lossy files I can safely conclude I prefer 56_12 because it's sounds closer to the flac, the 56_13 contains more distortion, especially in the 0:03.3-0:04.9 range. I tested with headphones.

CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9 beta 1
2009/08/13 19:20:26

File A: F:\Temp\muaddib\emese.flac
File B: F:\Temp\muaddib\emese_56_12.mp4

19:20:26 : Test started.
19:21:10 : 01/01 50.0%
19:21:21 : 02/02 25.0%
19:21:33 : 03/03 12.5%
19:21:45 : 04/04 6.3%
19:21:57 : 05/05 3.1%
19:22:14 : 06/06 1.6%
19:22:28 : 07/07 0.8%
19:22:43 : 08/08 0.4%
19:22:56 : 09/09 0.2%
19:23:11 : 10/10 0.1%
19:23:21 : 11/11 0.0%
19:23:26 : 12/12 0.0%
19:23:35 : 13/13 0.0%
19:23:46 : 14/14 0.0%
19:23:57 : 15/15 0.0%
19:24:13 : Test finished.

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


CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9 beta 1
2009/08/13 21:42:04

File A: F:\Temp\muaddib\emese.flac
File B: F:\Temp\muaddib\emese_56_13.mp4

21:42:04 : Test started.
21:42:40 : 01/01 50.0%
21:42:48 : 02/02 25.0%
21:42:58 : 03/03 12.5%
21:43:05 : 04/04 6.3%
21:43:11 : 05/05 3.1%
21:43:17 : 06/06 1.6%
21:43:23 : 07/07 0.8%
21:43:28 : 08/08 0.4%
21:43:35 : 09/09 0.2%
21:43:42 : 10/10 0.1%
21:43:50 : 11/11 0.0%
21:43:59 : 12/12 0.0%
21:44:09 : 13/13 0.0%
21:44:18 : 14/14 0.0%
21:44:23 : 15/15 0.0%
21:44:32 : Test finished.

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


CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9 beta 1
2009/08/13 21:48:47

File A: F:\Temp\muaddib\emese_56_12.mp4
File B: F:\Temp\muaddib\emese_56_13.mp4

21:48:47 : Test started.
21:49:10 : 01/01 50.0%
21:50:16 : 02/02 25.0%
21:50:38 : 03/03 12.5%
21:50:53 : 03/04 31.3%
21:51:18 : 04/05 18.8%
21:51:35 : 05/06 10.9%
21:51:55 : 06/07 6.3%
21:52:07 : 07/08 3.5%
21:53:16 : 08/09 2.0%
21:53:26 : 09/10 1.1%
21:53:38 : 10/11 0.6%
21:53:45 : 10/12 1.9%
21:54:22 : 11/13 1.1%
21:54:36 : 12/14 0.6%
21:54:53 : 13/15 0.4%
21:54:55 : Test finished.

----------
Total: 13/15 (0.4%)

ABX-ing 56_12 against 56_13 was tough although 13/15 in the first attempt suggests the opposite.

This post has been edited by Alexxander: Aug 13 2009, 21:23
Go to the top of the page
+Quote Post
Alexxander
post Aug 13 2009, 22:07
Post #14





Group: Members
Posts: 457
Joined: 15-November 04
Member No.: 18143



ABX-ing lossless bibilolo against lossy ones was easy with the tip given in other posts: crank up the volume and you can hear the background noise. Curiously (at least to me) the lossy samples have a lot less noise. I've noticed also a sound mainly on left side, right at the beginning of the track which I could also abx (tried just a few times) but the background noise was a lot easier to abx. I focussed on range 3.0-6.1 sec:

CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9 beta 1
2009/08/13 22:30:55

File A: F:\Temp\muaddib\bibilolo.flac
File B: F:\Temp\muaddib\bibilolo_56_12.mp4

22:30:55 : Test started.
22:31:32 : 01/01 50.0%
22:31:36 : 02/02 25.0%
22:31:39 : 03/03 12.5%
22:31:43 : 04/04 6.3%
22:31:47 : 05/05 3.1%
22:31:53 : 06/06 1.6%
22:32:00 : 07/07 0.8%
22:32:07 : 08/08 0.4%
22:32:16 : 09/09 0.2%
22:32:25 : 10/10 0.1%
22:32:33 : 11/11 0.0%
22:32:43 : 12/12 0.0%
22:32:47 : 13/13 0.0%
22:32:56 : 14/14 0.0%
22:33:00 : 15/15 0.0%
22:33:02 : Test finished.

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


CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9 beta 1
2009/08/13 22:33:29

File A: F:\Temp\muaddib\bibilolo.flac
File B: F:\Temp\muaddib\bibilolo_56_13.mp4

22:33:29 : Test started.
22:34:07 : 01/01 50.0%
22:34:10 : 02/02 25.0%
22:34:15 : 03/03 12.5%
22:34:19 : 04/04 6.3%
22:34:23 : 05/05 3.1%
22:34:27 : 06/06 1.6%
22:34:30 : 07/07 0.8%
22:34:33 : 08/08 0.4%
22:34:38 : 09/09 0.2%
22:34:43 : 10/10 0.1%
22:34:49 : 11/11 0.0%
22:34:54 : 12/12 0.0%
22:34:57 : 13/13 0.0%
22:35:02 : 14/14 0.0%
22:35:11 : 15/15 0.0%
22:35:12 : Test finished.

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


ABX-ing 56_12 against 56_13 based on background noise was too hard, I gave up after countless tries so I listened to a foreground sound in the same range and this leaded to:
CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9 beta 1
2009/08/13 22:47:28

File A: F:\Temp\muaddib\bibilolo_56_12.mp4
File B: F:\Temp\muaddib\bibilolo_56_13.mp4

22:47:28 : Test started.
22:48:16 : 01/01 50.0%
22:48:30 : 02/02 25.0%
22:48:44 : 02/03 50.0%
22:48:54 : 02/04 68.8%
22:49:01 : 03/05 50.0%
22:49:11 : 04/06 34.4%
22:49:24 : 05/07 22.7%
22:49:33 : 06/08 14.5%
22:49:53 : 07/09 9.0%
22:50:10 : 08/10 5.5%
22:50:42 : 09/11 3.3%
22:51:24 : 10/12 1.9%
22:51:35 : 10/13 4.6%
22:51:51 : 10/14 9.0%
22:52:06 : 11/15 5.9%
22:52:12 : Test finished.

----------
Total: 11/15 (5.9%)

I'm sure I have found what to listen to (high frequency distortion) but my hearing seem to be fatigue, I hear tones with amplifier powered off wacko.gif I will give the latter one another try later this week.
Go to the top of the page
+Quote Post
muaddib
post Aug 14 2009, 09:17
Post #15





Group: Developer
Posts: 398
Joined: 14-October 01
Member No.: 289



Thank you IgorC and Alexxander.

QUOTE (Alexxander @ Aug 13 2009, 23:07) *
I'm sure I have found what to listen to (high frequency distortion) but my hearing seem to be fatigue, I hear tones with amplifier powered off wacko.gif I will give the latter one another try later this week.

Please don't hurt your ears with too loud volume. I believe that your results so far are enough. It would be good to get more results from other people.
Go to the top of the page
+Quote Post
IgorC
post Aug 25 2009, 01:46
Post #16





Group: Members
Posts: 1554
Joined: 3-January 05
From: ARG/RUS
Member No.: 18803



Well, not much results until now.
It will be usefull if some of AAC's users (maybe /mnt, kornchild ...) will participate here also.

@muaddib
If you want you can post more encoded samples. Maybe Alexxander and me can do some more test.

This post has been edited by IgorC: Aug 25 2009, 01:46
Go to the top of the page
+Quote Post
/mnt
post Aug 25 2009, 02:38
Post #17





Group: Members
Posts: 697
Joined: 22-April 06
Member No.: 29877



I've tried the emese sample, but i have to admit am not that great at ABXing it. Since its seems to be a ringing problem IMO. Which is a artifact i seem to be deaf to, but Nero AAC seems to a few ringing problems at 0.50 with my ears though.

56 12
CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9
2009/08/25 02:10:02

File A: C:\Downloads\56_12_vs_56_13\emese.flac
File B: C:\Downloads\56_12_vs_56_13\emese_56_12.mp4

02:10:02 : Test started.
02:10:15 : 01/01 50.0%
02:10:23 : 02/02 25.0%
02:10:36 : 03/03 12.5%
02:10:51 : 04/04 6.3%
02:11:05 : 05/05 3.1%
02:11:21 : 06/06 1.6%
02:11:31 : 07/07 0.8%
02:11:54 : 08/08 0.4%
02:12:13 : 09/09 0.2%
02:12:17 : 10/10 0.1%
02:12:18 : Test finished.

----------
Total: 10/10 (0.1%)


Metallic ringing like artifacts, i targeted on 0:02 - 0:06.

56 13


CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9
2009/08/25 02:14:51

File A: C:\Downloads\56_12_vs_56_13\emese.flac
File B: C:\Downloads\56_12_vs_56_13\emese_56_13.mp4

02:14:51 : Test started.
02:15:02 : 01/01 50.0%
02:15:06 : 02/02 25.0%
02:15:10 : 03/03 12.5%
02:15:18 : 04/04 6.3%
02:15:24 : 04/05 18.8%
02:15:34 : 05/06 10.9%
02:15:39 : 06/07 6.3%
02:15:46 : 07/08 3.5%
02:15:53 : 08/09 2.0%
02:15:58 : 09/10 1.1%
02:16:02 : 10/11 0.6%
02:16:07 : 11/12 0.3%
02:16:12 : 12/13 0.2%
02:16:20 : 13/14 0.1%
02:16:34 : 14/15 0.0%
02:16:41 : Test finished.

----------
Total: 14/15 (0.0%)


Same problem.

56 12 vs 56 13

CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9
2009/08/25 02:16:59

File A: C:\Downloads\56_12_vs_56_13\emese_56_12.mp4
File B: C:\Downloads\56_12_vs_56_13\emese_56_13.mp4

02:16:59 : Test started.
02:17:19 : 01/01 50.0%
02:17:28 : 02/02 25.0%
02:17:48 : 02/03 50.0%
02:18:03 : 02/04 68.8%
02:18:08 : 02/05 81.3%
02:18:14 : 03/06 65.6%
02:18:23 : 04/07 50.0%
02:18:35 : 05/08 36.3%
02:18:49 : 05/09 50.0%
02:18:58 : 06/10 37.7%
02:19:16 : 06/11 50.0%
02:19:20 : 06/12 61.3%
02:19:25 : 07/13 50.0%
02:19:31 : 08/14 39.5%
02:19:39 : 09/15 30.4%
02:20:11 : 09/16 40.2%
02:20:30 : 09/17 50.0%
02:20:33 : 09/18 59.3%
02:20:45 : Test finished.

----------
Total: 9/18 (59.3%)


At first i thought file A had disortion at 0:03 and B sounded 'faster', but gave up at the end.

I will try the bibilolo sample at a later time, after i had some sleep.


--------------------
"I never thought I'd see this much candy in one mission!"
Go to the top of the page
+Quote Post
IgorC
post Aug 25 2009, 04:14
Post #18





Group: Members
Posts: 1554
Joined: 3-January 05
From: ARG/RUS
Member No.: 18803



I tried emese sample few times more but results and preferences are the same.
Very close but...
CODE
ABC/HR Version 1.1 beta 2, 18 June 2004
Testname:

1L = C:\0\NERO Developres\emeses\emese_56_12.wav
2R = C:\0\NERO Developres\emeses\emese_56_13.wav

---------------------------------------
General Comments:

---------------------------------------
1L File: C:\0\NERO Developres\emeses\emese_56_12.wav
1L Rating: 4.5
1L Comment:
---------------------------------------
2R File: C:\0\NERO Developres\emeses\emese_56_13.wav
2R Rating: 4.3
2R Comment:
---------------------------------------
ABX Results:
C:\0\NERO Developres\emeses\emese_56_12.wav vs C:\0\NERO Developres\emeses\emese_56_13.wav
    4 out of 5, pval = 0.188


This post has been edited by IgorC: Aug 25 2009, 04:19
Go to the top of the page
+Quote Post
muaddib
post Aug 25 2009, 09:23
Post #19





Group: Developer
Posts: 398
Joined: 14-October 01
Member No.: 289



QUOTE (IgorC @ Aug 25 2009, 02:46) *
@muaddib
If you want you can post more encoded samples. Maybe Alexxander and me can do some more test.

There are not many samples that have artifacts at abr 224 kbps and it is expected that there are even less samples that could give noticeably different results when using two setting that are tested here.
So far the results were as expected.
It would be good if one or two more golden ears contribute.
Go to the top of the page
+Quote Post
Alexxander
post Aug 25 2009, 09:57
Post #20





Group: Members
Posts: 457
Joined: 15-November 04
Member No.: 18143



Pet Shop Boys - In the Night has problem spots with nearly all Lame 3.98 settings and is easy to ABX with Nero v1.3.3.0 q0.50. After some trying at q0.60 I'm not sure to be able to ABX consistently but if you want create encoded samples and I'll spend time ABX-ing.
Go to the top of the page
+Quote Post
/mnt
post Aug 25 2009, 11:36
Post #21





Group: Members
Posts: 697
Joined: 22-April 06
Member No.: 29877



Also Closer to God by NIN is very easy to ABX at q 0.60.

Anyway i tried the bibilolo sample:

56 12

CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9
2009/08/25 10:58:30

File A: C:\Downloads\56_12_vs_56_13\bibilolo.flac
File B: C:\Downloads\56_12_vs_56_13\bibilolo_56_12.mp4

10:58:30 : Test started.
10:58:50 : 01/01 50.0%
10:59:05 : 02/02 25.0%
10:59:18 : 03/03 12.5%
10:59:31 : 04/04 6.3%
10:59:42 : 05/05 3.1%
10:59:57 : 06/06 1.6%
11:00:06 : 07/07 0.8%
11:00:36 : 08/08 0.4%
11:00:46 : 09/09 0.2%
11:00:53 : 10/10 0.1%
11:00:54 : Test finished.

----------
Total: 10/10 (0.1%)


Metallic hiss around 0:08 - 0:10. Seems to share the same problem that a older version of Nero AAC had with silent parts on analog recordings.

56 13

CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9
2009/08/25 11:10:03

File A: C:\Downloads\56_12_vs_56_13\bibilolo.flac
File B: C:\Downloads\56_12_vs_56_13\bibilolo_56_13.mp4

11:10:03 : Test started.
11:10:19 : 01/01 50.0%
11:10:39 : 02/02 25.0%
11:10:56 : 03/03 12.5%
11:11:05 : 04/04 6.3%
11:11:20 : 05/05 3.1%
11:11:46 : 06/06 1.6%
11:12:07 : 07/07 0.8%
11:12:36 : 07/08 3.5%
11:12:42 : 07/09 9.0%
11:12:54 : 08/10 5.5%
11:12:59 : 09/11 3.3%
11:13:10 : 10/12 1.9%
11:13:27 : 11/13 1.1%
11:13:43 : 12/14 0.6%
11:14:00 : 13/15 0.4%
11:14:03 : Test finished.

----------
Total: 13/15 (0.4%)


Same metallic hiss problem. This sample is really prone to hearing fatigue.

56 12 vs 56 13

CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9
2009/08/25 11:16:34

File A: C:\Downloads\56_12_vs_56_13\bibilolo_56_12.mp4
File B: C:\Downloads\56_12_vs_56_13\bibilolo_56_13.mp4

11:16:34 : Test started.
11:16:54 : 00/01 100.0%
11:17:13 : 01/02 75.0%
11:17:30 : 01/03 87.5%
11:17:34 : 01/04 93.8%
11:17:40 : 02/05 81.3%
11:17:44 : 03/06 65.6%
11:18:05 : 04/07 50.0%
11:18:23 : 05/08 36.3%
11:18:32 : 05/09 50.0%
11:18:36 : 06/10 37.7%
11:18:52 : 07/11 27.4%
11:18:58 : 07/12 38.7%
11:19:03 : 08/13 29.1%
11:19:12 : 09/14 21.2%
11:19:15 : 10/15 15.1%
11:19:31 : Test finished.

----------
Total: 10/15 (15.1%)


Am not really sure which one is better.

This post has been edited by /mnt: Aug 25 2009, 12:30


--------------------
"I never thought I'd see this much candy in one mission!"
Go to the top of the page
+Quote Post
muaddib
post Aug 25 2009, 15:50
Post #22





Group: Developer
Posts: 398
Joined: 14-October 01
Member No.: 289



Here are two versions of the samples requested by Alexxander and /mnt.
Enjoy and thanks!
Attached File(s)
Attached File  56_12_vs_56_13__second_try_.zip ( 2.43MB ) Number of downloads: 221
 
Go to the top of the page
+Quote Post
Alexxander
post Aug 31 2009, 22:22
Post #23





Group: Members
Posts: 457
Joined: 15-November 04
Member No.: 18143



First I started to ABX 01___Pet_Shop_Boys___In_The_Night___cutted_12.mp4 against FLAC. I tried my known spots and couldn't distinguish any. Then for about 20 minutes I tried to find other spots with differences. Again I couldn't.

Then I went to ABX flac against 01___Pet_Shop_Boys___In_The_Night___cutted_13.mp4. One of my usual spots, this one between 9.9 and 10.9 secs, was ABX-able, more or less:

CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9
2009/08/31 22:58:26

File A: F:\Temp\muaddib\01 - Pet Shop Boys - In The Night --cutted.flac
File B: F:\Temp\muaddib\01___Pet_Shop_Boys___In_The_Night___cutted_13.mp4

22:58:26 : Test started.
22:58:51 : 00/01 100.0%
22:59:37 : 01/02 75.0%
22:59:51 : 02/03 50.0%
23:00:02 : 03/04 31.3%
23:00:17 : 04/05 18.8%
23:00:38 : 05/06 10.9%
23:01:17 : 05/07 22.7%
23:01:25 : 06/08 14.5%
23:01:34 : 07/09 9.0%
23:01:49 : 07/10 17.2%
23:02:22 : 08/11 11.3%
23:02:38 : 09/12 7.3%
23:02:48 : 09/13 13.3%
23:02:55 : 10/14 9.0%
23:03:06 : 10/15 15.1%
23:03:20 : 11/16 10.5%
23:03:31 : 12/17 7.2%
23:03:45 : 13/18 4.8%
23:03:50 : 14/19 3.2%
23:03:59 : 15/20 2.1%
23:04:05 : Test finished.

----------
Total: 15/20 (2.1%)

With patience I suppose I could do better.

Then I went back to cutted_12.mp4 and tried to ABX the same range around 10.0 secs. I found cutted_12.mp4 impossible to ABX.
Go to the top of the page
+Quote Post
IgorC
post Sep 1 2009, 07:28
Post #24





Group: Members
Posts: 1554
Joined: 3-January 05
From: ARG/RUS
Member No.: 18803



I obtain very similar results. It's kind a strange because I didn't really ABXed it by hearing but by feeling. It's weird. I would scored something like 4.999... ( being 5 is max. score) There wasn't any artifact but I felt like encoded file had extremely slightly less amplitude on cowbell in right channel.
Sample 13 is abxed and I was already tired for sample 12. Will try later.

CODE
foo_abx 1.3.4 report
foobar2000 v0.9.6.9
2009/09/01 03:11:12

File A: C:\0\Nero Developer Samples\01___Pet_Shop_Boys___In_The_Night___cutted.flac
File B: C:\0\Nero Developer Samples\01___Pet_Shop_Boys___In_The_Night___cutted_13.mp4

03:11:12 : Test started.
03:12:15 : 01/01  50.0%
03:12:19 : 01/02  75.0%
03:12:23 : 02/03  50.0%
03:12:28 : 02/04  68.8%
03:12:31 : 03/05  50.0%
03:12:50 : 04/06  34.4%
03:12:55 : 05/07  22.7%
03:13:09 : 05/08  36.3%
03:13:15 : 05/09  50.0%
03:13:18 : 06/10  37.7%
03:13:22 : 07/11  27.4%
03:13:25 : 08/12  19.4%
03:13:29 : 09/13  13.3%
03:13:33 : 10/14  9.0%
03:13:37 : 11/15  5.9%
03:13:42 : 12/16  3.8%
03:13:46 : 12/17  7.2%
03:13:50 : 13/18  4.8%
03:14:15 : 14/19  3.2%
03:14:22 : 15/20  2.1%
03:15:10 : Test finished.

----------
Total: 15/20 (2.1%)


This post has been edited by IgorC: Sep 1 2009, 07:37
Go to the top of the page
+Quote Post
Alexxander
post Sep 1 2009, 09:25
Post #25





Group: Members
Posts: 457
Joined: 15-November 04
Member No.: 18143



QUOTE (IgorC @ Sep 1 2009, 08:28) *
I obtain very similar results. It's kind a strange because I didn't really ABXed it by hearing but by feeling. It's weird. I would scored something like 4.999... ( being 5 is max. score) There wasn't any artifact but I felt like encoded file had extremely slightly less amplitude on cowbell in right channel.
Sample 13 is abxed and I was already tired for sample 12. Will try later.

Yes, no artifact but a very light kind of sweeping stereo, very hard to detect and only when concentrated at 200% and with headphones. Obviously this is not a problem, the original source sound could have been exactly this but we know it isn't because we have the flac.

This post has been edited by Alexxander: Sep 1 2009, 09:26
Go to the top of the page
+Quote Post

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: 23rd August 2014 - 14:58