Choice of AAC codecs for listening test Options

[rjamorim]

Hello.

I'm creating this poll so users can choose what codecs to be features at the AAC test.

Three codecs are already decided: iTunes, Nero and Faac.

So, please vote on the remaining two AAC codecs. The anchor will be decided in another poll.

Thank-you.

Regards;

Roberto.

[Cey]

I vote for Winamp because it is *very* widely available. It has more relevancy than the others. The whole idea is to test encoders people use and/or have easy access to. Winamp fits that quite well.

For the other codec... I don't really care. I don't know anything about comaact so I can't comment on that. And I don't like Real, their programs or their company, so I can't really endorse that either. (And their program is currently beta, is the codec stable?) Make yourself happy....

Edit: For the record, I voted for 'other' because of those listed, I only see WinAmp as being widely important. I don't really care what the second codec would be.

If you end up not using WinAmp, then I don't care what the other codecs are.

This post has been edited by Cey: Feb 9 2004, 19:18

[mdmuir]

Winamp and Real, only because of pervasive availability for "joe average"

[guest0101]

Agree with mdmuir, WinAmp and Real to represent these commonly used programs by users out there.

[MGuti]

compaact and winamp: AFAIK real aac can't really be used outside of realplayer 10 unless you purchase gold. if i can't transcode to anything i want with it stay away!

[kl33per]

compaact, because (As far as I know) it's completely in-house devloped, which could potentially make for a very interesting comparison. Not overly worried about the other encoder, NCTU or Winamp would be my pick though.

[harashin]

I'd like to see a FhG's encoder in the test. I got iisMP4serv demo version last year, on fatboy, I found it's the best encoder I've tried at 128kbps. Hence I vote for other. I don't care another anyway.

Edit: You can download the sample here.

This post has been edited by harashin: Feb 9 2004, 05:55

[rjamorim]

Something worth mentioning (so that later people don't say I'm changing the rules to adapt the results to my fancy)

This is how results will be calculated: Each vote actually weights for two (except votes to the other choice, with explanation)

So, as I write this, we have:
(C = Compaact, W = Winamp, R = Real, N = NCTU)

CODE

          C   W   R   N
C and W   8   8
C and N   4           4
W and N       1       1
R and C   1       1
R and W       7   7
R and N           1   1
          -------------
Sum       13  16  9   6

What would mean Compaact and Winamp are going to be tested.

Votes going to "other" will be weighted after I read the posts of users that voted for other. So, if you voted for it, please make it clear at the body of your message and explain what would be your choice(s).

Regards;

Roberto.

This post has been edited by rjamorim: Feb 9 2004, 05:50

[saratoga]

What is winamp's decoder? Completely new or based on Dolby?

[bond]

looks like people want winamp in either way and the decision has to be made between real and compaact

i voted for real because they are going to start an own online music shop with aac in .ra container (according to first news, i hope they change this to .mp4) and own drm
as real is going to use 192kbps (claiming that this should be better quality than apples 128kbps) i am really interested in how their codec does compared to apple


damn normally i would have voted for winamp, real and compaact, but as rjamorim doesnt want to use one codec more (or drop the anchor/take faac as "potential" anchor)...

[WEST]

Rjamorim, please test Nero AAC Encoder 2.6.1.9

[blessingx]

Real because I think it could have a greater impact on the acceptance of AAC in the corp world. Plus their online store.

[tigre]

Winamp, Real and Compaact, no anchor instead.

The 0-5 ABC/HR scale is associated with terms anyway (Very Annoying ... Imperceptible) - and I know when something sounds e.g. 'Perceptible, but not annoying' to me, so anchors are quite useless IMHO.

Edit: I've voted 'Other', but if I'd had to choose 2, it would be Winamp and Compaact.

This post has been edited by tigre: Feb 9 2004, 11:23

[unmake]

Is it more important to test encoders which might supposedly be used widely by other people, or those which might be widely used by HA members? Seems to me like anyone with half a brain cell in their head would steer clear of anything related to Real.

Then again, if their software does turn out to be up to snuff, I'd much rather have someone else installing it on their computer in order to make that discovery.

[kotrtim]

Null Vote
I suggest testing all the widely available codecs and also all the worth testing codecs

EDIT: I'm a bit confuse now, widely available and wide used, is it the same thing?????

add a few wont harm right

bcoz currently there are only 3 types on the list
adding 3 or 4 more wont hurt?

exercising the eardrum will not damage it

just my op

This post has been edited by kotrtim: Feb 9 2004, 13:17

[p0l1m0rph1c]

I'd choose FhG's codec and compaact. Hence i voted other.

[rjamorim]

Hello. Menno just shouted at me on ICQ:

m&no (05:36 AM) :
DON'T ADD WINAMP TO THE AAC TEST BEFORE THEY FIX THEIR BUGS!!!!
m&no (05:36 AM) :
their TNS is broken, at least in their decoder, but prolly they have matched bugs in encoder and decoder
m&no (05:37 AM) :
meaning that if you use a correct decoder on their files Winamp will completely loose this test
m&no (05:37 AM) :
which is not very fair, Winamp is prolly going to fix this soon
m&no (06:31 AM) :
hmm, maybe it's not that bad
Leviathan (11:53 AM) :
Well, OK...

So, I think it's a wise idea to give up Winamp? I would personally love to test it, but if it's unfair in the end...

[bond]

maybe someone who has a clue about that should test whether this bug is also in their encoder (i guess their decoder wouldnt have been used in the test anyways)

i would love to know how winamp does

[rjamorim]

I'm creating two test sets, one with AAC files encoded by Winamp and decoded by Winamp, other with files encoded by Winamp and decoded by Faad2.

[rjamorim]

OK, I created a package. Samples are encoded with Winamp, then decoded with Winamp or Faad2.
http://rarewares.hydrogenaudio.org/samples/quicktest.rar (26.6Mb)

No need for very fancy testing methods, just somehow compare _1 to _2 and post if you find any noticeable difference and, in that case, what sounded better.

Thanks a lot for your help.

Regards;

Roberto.

PS: Why RAR instead of ZIP? Because RAR compresses audio files much better. You can get freeware decompressors for nearly every platform here.

This post has been edited by rjamorim: Feb 9 2004, 16:08

[tigre]

Posts by brucewillis violating TOS #2 and #8 are split here.

[Alexander Lerch]

No need for very fancy testing methods, just somehow compare _1 to _2 and post if you find any noticeable difference and, in that case, what sounded better.

I had only a quick look at that files; version 1 seems often to clip. That might be the (one?) reason for differences in the high frequency range.

Alexander

[john33]

You don't ask for scientific evaluation, so, 2 sounds much better to me.

[tigre]

ABC/HR results

ABC/HR Version 0.9b, 30 August 2002
Testname: DaFunk

1L = N:\DaFunk_1.wav
2R = N:\DaFunk_2.wav

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

---------------------------------------
1L File: N:\DaFunk_1.wav
1L Rating: 3.5
1L Comment: 0.0-3.3: flanging, clackling noise, difference gets easily lost during quick ABXing
3.3-16.0: differences less obvious/annoying
16.0-20.6: differences similar to the beginning
---------------------------------------
2R File: N:\DaFunk_2.wav
2R Rating: 2.5
2R Comment: 0.0-3.3: flanging, 'airy' bumping noises added, very obvious, annoying
3.3-16.0: differences less obvious/annoying
16.0-20.6: differences similar to the beginning
---------------------------------------
ABX Results:
Original vs N:\DaFunk_1.wav
    11 out of 15, pval = 0.059
Original vs N:\DaFunk_2.wav
    6 out of 6, pval = 0.016
N:\DaFunk_1.wav vs N:\DaFunk_2.wav
    6 out of 6, pval = 0.016

ABC/HR Version 0.9b, 30 August 2002
Testname: Gone

1L = N:\gone_2.wav
2L = N:\gone_1.wav

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

---------------------------------------
1L File: N:\gone_2.wav
1L Rating: 3.0
1L Comment: 2.5-5.0: background noise brighter + changed (bumping)
---------------------------------------
2L File: N:\gone_1.wav
2L Rating: 3.0
2L Comment: 2.5-5.0: similar to 1, there seems to be a difference, but hard to tell which is better
starting at 9.0 there's flanging (cymbal sounds), again different between 1 and 2 but impossible to tell which is better for me.
---------------------------------------
ABX Results:
Original vs N:\gone_2.wav
    6 out of 6, pval = 0.016
Original vs N:\gone_1.wav
    12 out of 16, pval = 0.038
N:\gone_2.wav vs N:\gone_1.wav
    6 out of 6, pval = 0.016

[JohnV]

We had a problem with a troll registering lots of fake accounts and voting certain choice in this poll. One of the examples is here.
Maybe Roberto should consider starting the poll again in order to get valid results.

I've disabled new registrations for now.

Edit. This kind of behaviour makes me think that maybe there should be encryption used when saving results from ABC/HR.
Seems that zealotry could be a problem here.

This post has been edited by JohnV: Feb 9 2004, 17:54