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: New problem sample (It's probably) (Read 7074 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

New problem sample (It's probably)

My first post, although I've been lurking here since 2002 (good, old days when it was fashionable to use musepack  ).
I think I found a new problem sample (1.16 standard). It's from Sting's "It's Probably Me". I ABXed first second. Interesting fact is that it's (I guess) pre-echo/noise problem. Could somebody confirm? And yes, I am aware that musepack is a dead format  .

You can find the sample in uploads section http://www.hydrogenaudio.org/forums/index....showtopic=56005

Here are the logs:

WinABX v0.42 test report
07/06/2007 21:19:04

A file: D:\07 - It's Probably Me.wav
B file: D:\Sting - It's Probably Me.wav

Start position 00:00.0, end position 00:01.5
21:21:19    0/1  p=100.0%
21:21:51    0/2  p=100.0%
21:21:54  reset

21:24:45    1/1  p=50.0%
21:25:13    2/2  p=25.0%
21:27:33    3/3  p=12.5%
21:28:44    4/4  p=6.2%
21:30:23    4/5  p=18.8%
21:33:55    5/6  p=10.9%
21:36:00    6/7  p=6.2%
21:40:14    7/8  p=3.5%
21:40:55  test finished


foo_abx 1.3.1 report
foobar2000 v0.9.4.1
2007/06/29 22:43:38

File A: D:\Muzyka\___Inne\07 - It's Probably Me.flac
File B: D:\07 - It's Probably Me.mpc

22:43:38 : Test started.
22:47:36 : 01/01  50.0%
22:48:32 : 01/02  75.0%
22:49:14 : 02/03  50.0%
22:50:49 : 03/04  31.3%
22:51:33 : 04/05  18.8%
22:54:32 : 05/06  10.9%
22:56:48 : 06/07  6.3%
22:59:20 : 07/08  3.5%
23:01:26 : 07/09  9.0%
23:02:27 : 07/10  17.2%
23:03:40 : 07/11  27.4%
23:04:09 : 07/12  38.7%
23:04:24 : 07/13  50.0%
23:04:28 : Test finished.

----------
Total: 7/13 (50.0%)

In the second test it's really 7/8, After 8th trial I couldn't believe my luck and clicked randomly.

New problem sample (It's probably)

Reply #1 –
Anyone?

I tried also:
- vorbis (aoTuV b5 q6) - much easier for me to distinguish
- mp3 (Lame 3.97 V2)

foo_abx 1.3.1 report
foobar2000 v0.9.4.1
2007/07/08 23:50:09

File A: D:\Probably_Me.ogg
File B: D:\Temp\Probably_Me.flac

23:50:09 : Test started.
23:50:29 : 01/01  50.0%
23:50:35 : 02/02  25.0%
23:50:45 : 03/03  12.5%
23:50:50 : 04/04  6.3%
23:51:05 : 05/05  3.1%
23:51:10 : 06/06  1.6%
23:51:20 : 07/07  0.8%
23:51:26 : 07/08  3.5%
23:51:43 : Test finished.

----------
Total: 7/8 (3.5%)

WinABX v0.42 test report
07/09/2007 22:14:17

A file: E:\Probably_Me_mp3_v2.wav
B file: E:\Probably_Me_flac.wav

Start position 00:00.0, end position 00:01.4
22:14:54    1/1  p=50.0%
22:31:37    2/2  p=25.0%
22:31:41    3/3  p=12.5%
22:31:50    4/4  p=6.2%
22:31:57    5/5  p=3.1%
22:32:00    6/6  p=1.6%
22:32:07    6/7  p=6.2%
22:32:10    7/8  p=3.5%
22:32:16  test finished

New problem sample (It's probably)

Reply #2 –
MPC differences are usually subtle at -standard. If there really is something there that's no big deal then --xtreme should get it out of your brain. Also you want better score on these subtle artifacts. Try 2 sets of 8 trials and maybe you can get < 3 %

New problem sample (It's probably)

Reply #3 –
LinABX

"12% correct (p=0.965)

You had 1 answer of 8 trials correct, which gives a correctness of 12%.
Compared to a coin-flip experiment, there is a chance of 0.965 that this happens. Given the fact that this is bigger than 0.050, this can be seen as a likely situation.
Given the fact that the number of trials is smaller than 16, these results may be highly inaccurate.

You are probably unable to hear the difference."


I kinda gave up after 4 steps cause i can't hear dif at all

Musepack -quality 5
<3

New problem sample (It's probably)

Reply #4 –
Quote from: shadowking on
MPC differences are usually subtle at -standard. If there really is something there that's no big deal then --xtreme should get it out of your brain. Also you want better score on these subtle artifacts. Try 2 sets of 8 trials and maybe you can get < 3 %


Of course it's subtle. All musepack problem samples are (for me) very difficult to distinguish. And I'm not worried about it at all, on the contrary I was happy when I found it  . I am not sure what you mean by "you want better score". I did two ABX test with musepack (only tests I did) - each of them 7/8 (fixed number of trials). Can you actually hear the problem? (for me it's the first "finger snapping", clicking sound  from 0,7 to 0,9s). Vorbis was much easier to ABX.


Quote from: mr budzone on
LinABX

"12% correct (p=0.965)

You had 1 answer of 8 trials correct, which gives a correctness of 12%.
Compared to a coin-flip experiment, there is a chance of 0.965 that this happens. Given the fact that this is bigger than 0.050, this can be seen as a likely situation.
Given the fact that the number of trials is smaller than 16, these results may be highly inaccurate.

You are probably unable to hear the difference."


I kinda gave up after 4 steps cause i can't hear dif at all

Musepack -quality 5
<3


Cool, thanks for trying. I'm usually bad at ABXing  that's why I asked for confirmation. 
 

New problem sample (It's probably)

Reply #5 –
In the beginning I can't hear a thing and hit 4/8.. then 7/8, 6/8 total 17/24

On the 2nd and 3rd attempts its as if there is a distorsion like a bump on the bass when the click happens. In my experience there are others like these subtle effects at --standard that you might abx with some attention, but not on --xtreme.