Choice of anchor/6th codec for AAC test Options

[rjamorim]

Hello.

Let's see if this time noone attempts to mess the poll >_<

OK, let's choose an anchor - or a 6th competitor.

When choosing, keep in mind the test will be already quite fatiguing as it is - several codecs that probably come close to transparency at this bitrate. So, I would probably recommend a bad quality anchor, I believe such would considerably reduce fatigue.

Of course, keep that even more in mind if you are planning to vote on a 6th competitor :B

Quoting the master

the anchor

1) should be significantly worse than any of the other codecs.

2) should sound bad enough to be readily identifiable. Yes, that will compress the ratings, but it will take a lot of the burden off the listener, who now need concentrate on one less codec. Also, it may encourage more people to participate, who might otherwise be scared off. I think lame and audioactive are too good, but the other mp3 codecs might work.

Also, keep in mind Lame has already been tested a lot: first multiformat test, 64kbps test, MP3 test... and it will also be tested at the next multiformat test, since it won the MP3 test.

Thanks for your help.

Regards;

Me.

***EDIT***: I don't guarantee the winner will be featured in the test. I will also depend on discussions and opinions.

Also, I do guarantee that if anyone messes with this poll, I'll bring everything down, force my choice of codecs down everybody's throat and God help us all! Don't start me, punks

This post has been edited by rjamorim: Feb 10 2004, 05:14

[guest0101]

My vote is for the 6th competitor (WinAmp) which got the most votes in the previous poll and beat out the "two winners" of the poll. This is likely the only Dolby based one and should be relatively un-tweaked for performance giving less than stellar performance and serve to be a good "anchor".

This post has been edited by guest0101: Feb 10 2004, 05:16

[music_man_mpc]

I think it would be funny to have the very first mp3 encoder in a test with all the latest and greatest AAC encoders.

[rjamorim]

This is likely the only Dolby based one

QuickTime is based on Dolby (with further tunning done inside Apple)

should be relatively un-tweaked for performance

Dude, Dolby knows quite a lot about audio coding. I would guess they tuned their encoder very well.

This post has been edited by rjamorim: Feb 10 2004, 05:17

[guest0101]

Are you just going to throw the WinAmp encoder out then, even though it "won" the poll? It had more votes than any others (before the vote tampering) if I recall correctly.

It would be nice to see the most popular programs included in the test:

1. iTunes (AAC)
2. Nero
3. WinAmp
4. Compaact!
5. RealPlayer
6. FAAC

But I know very little about your testing procedures, so if you must, then pick another one for an "anchor".

This post has been edited by guest0101: Feb 10 2004, 05:21

[rjamorim]

Are you just going to throw the WinAmp encoder out then, even though it "won" the poll?

Nope. I was considering throwing it out because it's broken, but nobody is completely sure about it yet, it seems.

That poll was tainted, so I won't base my choice of codecs on it at all.

And no, there has been no decision yet to throw Winamp out. First I need to know about the brokenness, and then I need to know about people's opinion.

Also, have you read the point about fatigue?

This post has been edited by rjamorim: Feb 10 2004, 05:22

[guest0101]

I thought Menno addressed that in the other thread. He said the WinAmp decoder was broken and not the encoder...

Anyway I trust your judgment, so do as you deem is the best for the test.

This post has been edited by guest0101: Feb 10 2004, 05:23

[rjamorim]

I thought Menno addressed that in the other thread. He said the WinAmp decoder was broken and not the encoder...

This is a huge mess. Menno also said "don't take my word for it" earlier. (he deleted this post later)

Also, few people cared to take my quick test and compare FAAD2 decodes vs. Winamp decodes of Winamp streams.

http://www.hydrogenaudio.org/forums/index....ndpost&p=182595

Also, the few results submitted make my head spin. Alexander and John33 claim encodes _2 sound better. Well, let me tell you, _2 was decoded with the Winamp decoder (that is supposedly broken)

This post has been edited by rjamorim: Feb 10 2004, 05:30

[ScorLibran]

I voted for an older FAAC version as an anchor, and at a low enough bitrate to be effective in this capacity, of course.


I'd also like to see Winamp as a competitor, IF it can be fixed by next Tuesday's deadline, of course. Considering the size of it's userbase, including it could expand the interest in the results of this test.

[bidz]

LAME 3.95.1 - simply because i want to see how it does against the AAC encoders, not just the winning codec of the AAC test (in the multiformat test). Featuring the best MP3 encoder also makes sense, since the AAC format is kind of the successor to the MP3 format. This would be a way of comparing old versus new.

I Would also like to se Winamp 5.02 featured in this test. This makes very good sense, considering it's userbase. I bet there are alot more users already encoding with Winamp 5.xx AAC than with Compaact, FAAC and NCTU-AAC (or what it's called) already, thus it deserves to be featured in the test (wether or not it's broken, that's a codec bug, and Nullsoft's problem in the end).

Not every codec is perfect, thus i don't see the reason as to why Winamp should be thrown out just because there MAY be a flaw with it. If there is a flaw with it, then let the test show that result. Maybe that would get Nullsoft to speed things up a bit

This post has been edited by bidz: Feb 10 2004, 05:53

[rjamorim]

and at a low enough bitrate to be effective in this capacity, of course.

Erm... rule nr. two, no mixed bitrates, pal

Unless you have some really good reason, like in the 64kbps test.

[Bonzi]

I voted for NCTU AAC I just really want to see this in a test. Anyway, Pystel would be interesting as well as we might be able to begin to get an idea how much better the Nero encoder is and also it would be interesting to see if the gap between Pystel and FAAC has closed at all from the last test that is.

[QuantumKnot]

I'll vote for Xing as an anchor, for the reasons I stated in the other thread.

[GodHead]

I voted for L3Enc because I used it a LOT in the past. I was in one of the original audio groups back in '96 that released some how-to's for L3Enc and WinPlay 3. I remember waiting ~hour for one MP3 encode and being amazed at the quality. I'd love to see what this old wonder can do in comparison to the new codecs.

@roberto: While browsing for possible anchors other than L3Enc, I came across the following, http://www.rjamorim.com/rrw/km/audiocomp1.htm. I never did mess around with the K+K stuff, what happened with it? I did a Google and noticed this page, http://www.cadaudio.dk/kk_research/. Never knew they had an AAC codec.

[ScorLibran]

and at a low enough bitrate to be effective in this capacity, of course.

Erm... rule nr. two, no mixed bitrates, pal

Unless you have some really good reason, like in the 64kbps test.

Oh, my bad. I didn't know that applied to the anchor, too.


(....Guess I'm having bad memories of rating the anchor higher than some of the competitors in the 64k test. )

[rjamorim]

@roberto: While browsing for possible anchors other than L3Enc, I came across the following, http://www.rjamorim.com/rrw/km/audiocomp1.htm. I never did mess around with the K+K stuff, what happened with it?

O_o

You found the mirrored page but didn't find the page linking to it?
http://www.rjamorim.com/rrw/tac.html

K+K history is available there.

BTW, I know that page. Check the domain, or the copyright notice at the bottom of the page I linked

I did a Google and noticed this page, http://www.cadaudio.dk/kk_research/. Never knew they had an AAC codec.

They hadn't

Read that page I linked.

This post has been edited by rjamorim: Feb 10 2004, 06:57

[ff123]

Is it possible to make some samples available to compare so that we can tell if the anchor will be bad enough?

ff123

[rjamorim]

  Oh, my bad.  I didn't know that applied to the anchor, too.

It's all a matter of purpose. If your purpose is trying to actually compare the anchor to the competitors, bitrates should be the same.

If it's only there to avoid codecs dipping too low, any bitrate will do.

In the 64kbps test, it was there specifically to check the validity of the "same quality as MP3 at half bitrates" touted by some codecs (WMA, MP3pro, Nero)

[rjamorim]

Is it possible to make some samples available to compare so that we can tell if the anchor will be bad enough?

Sure.

You mean l3enc only, or all suggested anchors?

[ViPER1313]

I like the idea of Lame v3.95.1 as an anchor, not because it is low quality, but because that would give a direct comparsion of the best of MP3 vs. all of AAC.

[bond]

i voted for leaving the anchor out, cause only this way we can test winamp, compaact and real!
otherwise we would have to leave one out...

[knik]

Definitely Lame here. It has many advantages:
It can act as a inter-test anchor.
It's very popular so the comparison to many AAC encoders would be very interesting.
Finally, all those AAC encoders have very high quality @128kbps. In the first AAC test even faac1.17 got 3.52 without any anchor.
Latest AAC encoders are supposed to be even better so a low quality anchor could boost all results well above 4.0 which is not the right thing.
LAME really looks like a perfect anchor to me. It wouldn't flatten results too much.

[menno]

This is a huge mess. Menno also said "don't take my word for it" earlier. (he deleted this post later)

That was before I was exactly sure what the problem was. The Winamp encoder is just as safe to use as any other encoder, I already said it before. The Winamp decoder has a problem.

Menno

[Alexander Lerch]

Although I know this will the test make much more difficult, I also voted for leaving the anchor out. I think Winamp should be tested.

Alexander

[JohnV]

I also voted: no anchor, take WinAmp.

But, another thing is also, that with 6 AAC codecs, the tested samples must be hard enough in order to achieve any kind of statistical difference between the codecs..