64kbps listening test, OPEN Options

[pseudoacoustic]

would you please do a quick abx test?

BadHorsie

..... to wav files, right? I'm all for helping you out but 1.) Wavs, even losslessly compressed, are HUGE, and 2.) I don't think I'd be able to tell the difference. They are.... um.... the same, and all (unless you have a bad CD drive, etc.)

Unless of course you're talking about a compressed format.

[BadHorsie]

try to rip a song a few times and do a md5sum. you will be suprised.
try to rip a cd which are pre-emphased with eac and cdparanoia. you will be suprised.

as far as i know had no one ever abx'ed eac vs. cdparanoia rips. i think it could not be bad to do it once.

BadHorsie

This post has been edited by BadHorsie: Sep 12 2003, 02:18

[rjamorim]

I don't see much point in ABXing EAC vs. CDex. If a CD is clean they will sound identical. If it's scratched or dirty, EAC will fail on some parts, CDex will fail on others. It's unlikely that only one of them will fail, unless it's a slight error to which one of them is prepared and the other isn't.

MD5sums and binary comparations also don't make sense at all. All you need is an offset difference of one bit for the MD5s to be completely different.

Heck, you can even get different sums if you rip the same track twice with any of these rippers. Even if the CD is perfectly clean.

Pre-emphasis means nothing either. Just because a ripper does de-emphasis while ripping, doesn't mean you can't run a de-emphasis filter on the ripped wave file after ripping. (You can use Sox for that)

Besides, since you migrated completely to Linux, there's not much of a choice. Either be happy with cdparanoia or go back to Windows.

Regards;

Me.

This post has been edited by rjamorim: Sep 12 2003, 02:36

[phong]

Shockingly enough, apparently EAC can be made to work under Wine, though I haven't had any success with it. Anyhow, I've got a script that rips CDs twice with cdparanoia to be secure that it does it right. So far, with the exception of some significantly damaged CDs the MD5 sum comes out the same ever time. Some drives might be less reliable (I've only tried the three I've got in my computer - an ultraplex, a plexwriter and a Nakamichi 5 disc changer).

[2Bdecided]

You don't ABX lossless audio - you compare bits, and that's that.

Time align, inverse mix paste in CEP is fine, though there are other methods.

If the two are not identical, then one is at fault. What it sounds like does not matter - lossless (e.g. ripping CD to .wav) should be a perfect copy of the original. If it's not, there's a fault.


Back to the original topic: How many samples do you want us to try rjamorim?

Everyone: try this test, it's quite good fun.

Cheers,
David.

[Mac]

I was going to say, please keep this thread about the 64k test.. this is something actually important


To Roberto, could you please plan a 32kb test next, at 64kb I'm finding that I can spot about half of the codecs..

As 2b sais, this test is a lot more enjoyable, it's a bit less demoralising now I can pick out at least 2 codecs every time

[bond]

anyone here to announce the test at slashdot?

[rjamorim]

Back to the original topic: How many samples do you want us to try rjamorim?

As many as you have time, patience and bandwidth to try.

If you can test all of them, that's perfect. But if you can only test one, that's already of much help to me.

BTW, talking about listening tests, would you be interested in my help for that MP2 listening test you were planning? If you want, we can discuss this over e-mail.

To Roberto, could you please plan a 32kb test next, at 64kb I'm finding that I can spot about half of the codecs..

Well, that's a possibility, of course. But I have other priorities first. Maybe I can conduce something like that early 2004

anyone here to announce the test at slashdot?

Thanks for reminding, that would indeed be welcome. Also, announcing at Kuro5hin and <shivers> r.a.o would be interesting.

Regards;

Roberto.

[lexor]

do you want us to submit the news to comunity sites we are registered with? like DeskMod? guys over there are getting pretty bored (now that they can't submit their app skins/visualizations/wallpapers etc...)

[bond]

why not?

the more people who join the test, the more work for rjamorim

[rjamorim]

Yes, the more it is announced, the better for all of us

the more people who join the test, the more work for rjamorim

Evil yu0 :B

[[proxima]]

I'm listening a sample and i occured with a possible bug in ABC-HR (this is not the 1st time). The samples are not played anymore, only the first part.
I've already done the comments for the samples and i don't want to restart the ABC-HR session again... any chance to recover the situation ? I'm afraid i should restart abc-hr

This post has been edited by [proxima]: Sep 13 2003, 16:42

[ancl]

,Sep 13 2003, 05:40 PM] I'm listening a sample and i occured with a possible bug in ABC-HR (this is not the 1st time). The samples are not played anymore, only the first part.

I've never encounted that error. Are you sure that you have not changed which part of the track that should be listened at? (The bar at the bottom of the screen...)

(Edit:Spelling)

This post has been edited by ancl: Sep 13 2003, 16:56

[[proxima]]

I've never encounted that error. Are you sure that you have not changed which part of the track that should be listened at? (The bar at the bottom of the screen...)

Thanks. You're right, i've played a little with the bar and the problem is now resolved. I don't know if there was a bug, maybe i'm just too tired... for today i think i'll stop doing listening test

[nwn]

Guru posted on other forum and I decide to come on the evaluation process

but .. as i try many investigations ..the fact is that i don't hear any sound from ABC ..

so let me explain config ..
what is the package i choose same issue so fix on the Sample01.zip

i decompress the file and all files are in the subdir \Sample01\
sure i've run the Sample01.bat (and not exe) .. and Sample01 files are converting in wav extension

all these wave are correct cause i can hear them in foobar.. good player isn'it ..

i launch abx and load config01 and see infos and buttons active (blue etc ..)

but which button and package (i try 5) no sound at all ..

observ .. portion selction bar is active but 0 / 0 0 sec displayed ...

here is the config01.txt
TestName = BigYellow Listening Test

Original = .\Sample01\BigYellow.wav
Sample1 = .\Sample01\BigYellow_1.wav
Sample2 = .\Sample01\BigYellow_2.wav
Sample3 = .\Sample01\BigYellow_3.wav
Sample4 = .\Sample01\BigYellow_4.wav
Sample5 = .\Sample01\BigYellow_5.wav
Sample6 = .\Sample01\BigYellow_6.wav
Sample7 = .\Sample01\BigYellow_7.wav
Sample8 = .\Sample01\BigYellow_8.wav

Offset_Orig = 14
Offset1 = 0
Offset2 = 14
Offset3 = 14
Offset4 = 18
Offset5 = 14
Offset6 = 48
Offset7 = 48
Offset8 = 14

___________________________________________________
subfolder

Directory of C:\Downloaded\Test_64_abc-hr\Sample01
bigyellow.wav bigyellow_1.wav bigyellow_2.wav bigyellow_3.wav bigyellow_4.wav bigyellow_5.wav bigyellow_6.wav bigyellow_7.wav bigyellow_8.wav
9 file(s) found
Total file size 35 815 584 bytes

[pseudoacoustic]

observ .. portion selction bar is active but 0 / 0 0 sec displayed ...

lol I think you've deselected what ABC-HR sould play

Just select the entire sample (highlight the entire bar so that it's shaded), and then hit a play button. The bar can be shortened(/deselected) so that only particular parts of the file are played.

[nwn]

i'm newbie for ever but not stupide .. no matter
it's a joke ..

the all bar is grayed

i read the posts....

yes i understand your analysis
thanks to help




This post has been edited by nwn: Sep 13 2003, 22:54

[pseudoacoustic]

i'm newbie for ever but not stupide .. no matter

Never said you were. But you did say 0/0 seconds, and that means nothing's selected, at least on my copy...

EDIT: Do you have a screenshot of abc-hr you can post?

This post has been edited by pseudoacoustic: Sep 13 2003, 22:44

[ff123]

From the screenshot it looks like everything's been loaded up properly, except that the portion select bar says 0.0 on both ends (it should say 30.0 sec at the right end for the "gone" sample).

I'm not sure what went wrong, but you could try setting it up manually by going to the File -> Setup Test menu and entering the files and the offsets by hand (if you loaded the config file properly, they should all be there already, though).

ff123

[nwn]

hi fff123 happy for your post ....

yes, i was surprise for "0" value too ... and conclude this is the problem ..

i've already look for ... and change manually path .. cause i thought taht ".\" was maybe not good .. ("..\" was maybe better)

it's what for i published the config01.txt .. cause in fact .. I dont't know if offset values are the rights one ....

but as i change manually the setup no more success ...

the config01.txt is not changed by manual setting?
as i press the Ok buttun soft ask for saving result test txt ....
but the config01.txt is not changed

version 0.9b 30/8/02

i looked for your site before .. see the explanation about error ...
"AAC decoder, FAAD (dated before about the early part of 2003) had a bug"
and didn't find any new version

okay .. that's all ..

no ...
all wave seems to be Pcm but lenght time are 24 sec except one is 25 sec ...
and curious suscpect .... some files haven't the same bitrate and bits

Sample01\BigYellow_4.wav
----------
540911 samples @ 22050Hz
bitrate = 705
samplerate = 22050
channels = 2
codec = PCM
bitspersample = 16
----------
540911 samples @ 22050Hz

Sample01\BigYellow_7.wav
bitrate = 1024
samplerate = 32000
channels = 2
codec = PCM
bitspersample = 16
----------
783360 samples @ 32000Hz

all others are

bitrate = 1411
samplerate = 44100
channels = 2
codec = PCM
bitspersample = 16
----------
1143808 samples @ 44100Hz

is there a problem with codec ... or calculate process ... i saw wich codec are generating this files ...

that's all for now ...
what can i do ..

This post has been edited by nwn: Sep 14 2003, 01:06

[ff123]

i've already look for ...  and change manually path .. cause i thought taht ".\" was maybe not good .. ("..\" was maybe better)

Try using the browse button to select the file.

it's what for i published the config01.txt .. cause in fact .. I dont't know if offset values are the rights one ....

Try setting up the test manually one sample at a time to see where the error occurs. I would use an offset of 0 for the samples until you get all 8 files loaded, then start changing the offsets one by one.

but as i change manually the setup no more success ...

the config01.txt is not changed by manual setting?
as i  press the Ok buttun soft ask for saving result test txt ....
but the config01.txt is not changed

Setting up the test manually does not change the config file. The results file is completely separate from the config file. The results file stores your comments and your ratings.

version  0.9b 30/8/02

i looked for your site before .. see the explanation about error ...
"AAC decoder, FAAD (dated before about the early part of 2003) had a bug"
and didn't find any new version

No new version was needed to fix this error -- FAAD was not reporting the WAV properties correctly, and the error was fixed with a recent version of FAAD2.

ff123

[nwn]

okay i did exactly what you said ...

i already use the browse button and lear insert value ...

but i just restart the process ...
neither the right time value changed ....
always "0" ....

ok ..

i must to try with another pack from another test ...
where can i found a pack which you are certain works ...



no move any file .... respect original tree and file position
hi 2bdecided

This post has been edited by nwn: Sep 14 2003, 01:23

[2Bdecided]

This is probably a stupid question, but you haven't moved abchr.exe have you? e.g. onto the desktop etc?

Cheers,
David.

[ff123]

okay i did exactly what you said ...

i already use the browse button and lear insert value ...

but i just restart the process ...
neither the right time value changed ....
always "0" ....

ok ..

i must to try with another pack from another test ...
where can i found a pack which you are certain works ...

Just to be clear:

You tried just the reference and one test file to start with? You can make the test file also point to the reference file (just in case one of the decoded test files are messed up, for some reason or another). Leave the offset at 0.

After you hit ok, the first group out of the eight will become active. The portion select bar should also have 30.0 sec on the right.

I'm unable to replicate this problem with the "gone" package, by the way.

ff123

[ff123]

I wonder if there's some problem with the different sample rates?

That's another reason to point to just the reference file, which is sampled at 44.1 kHz.

ff123

This post has been edited by ff123: Sep 14 2003, 01:30