Skip to main content

Topic: Pre-Test discussion (Read 32296 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
  • S_O
  • [*][*][*][*]
Pre-Test discussion
Reply #50
Quote
Heck, Atrac3 loses to other codecs (in features) in nearly every imaginable way (not multi platform (only Windows)), only CBR, no open decoder or encoder, not usable in media players like Winamp or Foobar, not usable in movies besides Real ones (unless you do hacks)...

This is not completly right:
-not multi-platform:
Everybody is allowed to create a en/decoder for every platform. Real will compile the libraries for every platform they said. This is part of the HelixCommuinty-project.
-Playable in every DirectShow supporting player thanks to Gabest Filters
-In movies it is useable, also thanks Gabest and it´s not a hack, Real has nothing against using their codecs in other containers. By the way, the new RealVideo 9 EHQ Video Codec has won in several tests (against XviD, WMV9, DivX Kauehi), I would like to know how the audio quality is. In the last month I got a new picture of Real, not the evil-format anymore.

And for wma9pro:
-Nearly no player, because no DirectShow support (only WMP7+???)
-Only windows
-No open-source en/decoder
-Not useable in any movies except wmv

For encoding I would use the Helix-Producer, because it´s free for everybody, not too big and creates unencrypted files. Also most people know Real (and maybe Helix), but who knows SonicStage and omg-files? If you ask Karl Lillevold (a Real developer) over Doom9 board, I think he will tell you exactly what kind/version etc. of Atrac3 they´ve used.

Quote
no legal way

I think it should be legal to create a winamp-plug-in if it´s legal to create DirectShow-Filters. You are not allowed to rip the codecs from RealOne, so the player must be installed, or you must use the binaries from the helix-project. But nobody has done a plug-in yet.

  • bond
  • [*][*][*][*][*]
Pre-Test discussion
Reply #51
Quote
By the way, the new RealVideo 9 EHQ Video Codec has won in several tests (against XviD, WMV9, DivX Kauehi), I would like to know how the audio quality is.

Me too
although i dont know any representative tests which prooves ehq to be better than xvid... (ok this is far too of topic)

Quote
I think it should be legal to create a winamp-plug-in if it´s legal to create DirectShow-Filters.

there is a plugin, but it was always said that it was illegal (although it needs realone installed like gabest's filter) dont know if the filter are legal or just tolerated (till now)
I know, that I know nothing (Socrates)

  • rjamorim
  • [*][*][*][*][*]
Pre-Test discussion
Reply #52
Here I come again.

@Dibrom: about the target audience, I think it'll be mixed. We'll have the main audience here at HA for sure, but we must also consider the <cough> guys at Slashdot that will also know about the results if things go as I hope.

Another option would be creating a test for widespread advertizing, that would appeal to the masses, and another one for HA mostly, that would focus on the codecs used at HA. This one I'll start next wednesday would be the "masses" one, because it already attracted quite some attention due to the AAC test. (including attention from Dolby labs and Apple  )

Anyone with thoughts on this idea? (of separating the test in two)

@Gecko: What I mean that is that JohnV votes for a different lowpass for each one of the 12 samples. E.G, ATrain gets a lowpass of 11kHz, and Waiting gets 9.5kHz. ff123 would vote for ATrain and Waiting (and everything else) using a fixed lowpass.

@Bond: There is a Real Audio plugin for Winamp called InnoReal. But it's "illegal" for some reason (Real demanded that Nullsoft remove it from their plugin database some years ago). I think that it's because they don't want people decoding their streams (what's stoopid, IMO).
Get up-to-date binaries of Lame, AAC, Vorbis and much more at RareWares:
http://www.rarewares.org

  • rjamorim
  • [*][*][*][*][*]
Pre-Test discussion
Reply #53
Quote
This is not completly right:
-not multi-platform:
Everybody is allowed to create a en/decoder for every platform. Real will compile the libraries for every platform they said. This is part of the HelixCommuinty-project.
-Playable in every DirectShow supporting player thanks to Gabest Filters
-In movies it is useable, also thanks Gabest and it´s not a hack, Real has nothing against using their codecs in other containers. By the way, the new RealVideo 9 EHQ Video Codec has won in several tests (against XviD, WMV9, DivX Kauehi), I would like to know how the audio quality is. In the last month I got a new picture of Real, not the evil-format anymore.

Hrm, I guess we have a problem here.

We need to define if, for our purposes, Atrac3 = RA or = Sony's version. That matters in the aspect that only one of these is usable in movies and playable in several environments, and only one is playable on hardware players. If both encoders are using the same encoding libraries, we can go ahead. But if they aren't, we'll have to chose between the multipurpose one of the hardware-playable one.

I'll mail Karl asking for clarification.

Regards;

Roberto.
Get up-to-date binaries of Lame, AAC, Vorbis and much more at RareWares:
http://www.rarewares.org

  • rjamorim
  • [*][*][*][*][*]
Pre-Test discussion
Reply #54
It's me again.

I just want to inform you that I'll travel to the middle of nowhere (for those that know about Brazilian geography: Mato Grosso) in some minutes, and I'm not sure yet if I'll be able to access the internet there. So, if I don't reply to this thread until next tuesday, fear not! - I'm still alive, and the test hopefully starts on time.

Best regards, and thank-you very much for everyone that helped so far.

Roberto.
Get up-to-date binaries of Lame, AAC, Vorbis and much more at RareWares:
http://www.rarewares.org

  • bond
  • [*][*][*][*][*]
Pre-Test discussion
Reply #55
Quote
And for wma9pro:
-Nearly no player, because no DirectShow support (only WMP7+???)
-Not useable in any movies except wmv

nope it is possible to mux wma9 pro into other containers (i tested it with gabest matroskamuxer (.mkv) and with m$ avi mux filter (.avi)) and to playback it with a m$ filter called "wmaudio decoder dmo" in directshow players

Quote
Anyone with thoughts on this idea? (of separating the test in two)

hm, so you would do 2 different tests (and in use both aac, vorbis and lame)? or one big "mass test" and the second one just a small test with some pros?

which codecs would participate in the mass test? aac, vorbis, lame, wma9 (std or pro?) [and realaudio or sony's atrac3 or real's cook  ]

Quote
@Bond: There is a Real Audio plugin for Winamp called InnoReal. But it's "illegal" for some reason (Real demanded that Nullsoft remove it from their plugin database some years ago). I think that it's because they don't want people decoding their streams (what's stoopid, IMO).

yup that was the plugin i meant, and that also was the reason why i am sceptical if real really "likes" gabest filter (although the should be wise enough to do so)...

edit: nice holiday rjamorim
  • Last Edit: 16 July, 2003, 07:42:15 PM by bond
I know, that I know nothing (Socrates)

  • JohnV
  • [*][*][*][*][*]
  • Developer
Pre-Test discussion
Reply #56
Quote
@Gecko: What I mean that is that JohnV votes for a different lowpass for each one of the 12 samples. E.G, ATrain gets a lowpass of 11kHz, and Waiting gets 9.5kHz. ff123 would vote for ATrain and Waiting (and everything else) using a fixed lowpass.

Actually, I meant that the results after lowpassing the anchor should be checked, because if using fixed lowpass, on some samples "the effect" can be much smaller than on some other samples. If there's a case where the chosen lowpass is clearly not enough to serve its purpose, it should be adjusted..
(I didn't give any values)
  • Last Edit: 18 July, 2003, 10:30:10 AM by JohnV
Juha Laaksonheimo

  • den
  • [*][*][*][*]
Pre-Test discussion
Reply #57
@Roberto and others

My understanding is that the Real version of ATRAC3 is not representative of the Sony ATRAC3 used in OpenMG and Sonicstage. For one, I don't think the same low pass is used.

Your best bet will be to either download a copy of SonicStage yourself, or get someone else to record the wavs on your behalf. If you go to www.minidisc.org and visit the various forums within, and then go to the NetMD sections, there is usually a sticky with links to downloads. Some of these are legal links, through Sony retailers, and some of them are warez.    You've just got to weave your way around all the 15 year olds posting "MD rulz!" 

Despite the plethora of samples, I feel that MPC and ATRAC3 are worth including. Many people on this board are not using mpc at 128 kbits, but how many have actually tried it?    Not many I suspect. As you mentioned, the few tests I've done have shown that mpc can be very competitive at 128 kbit... and I'm not alone with this impression. 

Can't wait for this test!

Den.

  • den
  • [*][*][*][*]
Pre-Test discussion
Reply #58
Sorry to double up, but there are also ways to play ATRAC3 .omg files in Winamp 2.81, or 2.95 (lite only) using a video plug in. The details are in one of the threads in the www.minidisc.org forums, but in essence you still need to have OpenMG or Sonicstage installed, and then you register the omgdec.ax file in Windows.

Once you have done that, you can either play the .omg file in Windows Media Player as is or Winamp with a plug in. Because it is still using the Sony omgdec file, I think you will find that the encryption restrictions still exist, ie you must have the original file present somewhere on you system, or have ripped it from an audio CD, but I haven't tested this to be sure.

If you already have Sonicstage/Open MG installed though, I can't see much benefit in doing this. My attempts to get Winamp to write a wav file to disk while playing an .omg file failed dismally. 

Den.

  • den
  • [*][*][*][*]
Pre-Test discussion
Reply #59
Me again... Last time this time, really... 

I just took Kraftwerks - The Robots and encoded it in Open MG (variant of Sonicstage), Real Producer and the crappy ATRAC3 acm codec. Encoded all three at 132 kbits using the standard "profile" in each case.

The ACM codec has a clear hi freq cut off at 17500 Hz.
Real Producer has a clear hi freq cut off at 16500 Hz.
Open MG also cuts off at 17500 Hz (edit: I was fooled before by OpenMG playing the wav instead of the omg file...>_<).

My personal vote is for Soundstage/OpenMG, but I have a vested interest. 
It is the one being pushed by Sony to replace LAME mp3 for portables, and is now being targetted by Apple and their I-pods (AAC), but it all depends what your target audience is for this test comparison.  B)

Den. 

Edit: Contained false info before regarding high freq cut off with Open MG.
  • Last Edit: 17 July, 2003, 12:54:24 AM by den

  • magic75
  • [*][*][*][*][*]
Pre-Test discussion
Reply #60
Quote
Quote
I vote both for keeping Musepack and for using WMA9pro.

OK... unless someone comes with good reasons to use WMA std or to sack Musepack, I'll go with your vote.

One reason from a newbie...
WMA std is supported on a lot of portables and WMApro is not?
Wasn't the use of Atrac in Sonys portables the initial reason for including it?
My personal opinion is that bitrates around 128 kbps is used typically on portables rather than played on a computer, and therefore formats supported by portables should be prioritized. That would mean dropping MPC and WMA9pro, but that's just me...

  • guruboolez
  • [*][*][*][*][*]
  • Members (Donating)
Pre-Test discussion
Reply #61
Short reply :

- Sorry, I haven't test BeautySlept with and without PNS encoding with musepack.
- Bachpsichord sample is available here :
http://membres.lycos.fr/guruboolez/AUDIO/samples/
- WMApro encodings are readable on foobar2000.
- I can confirm that winamp playback of encrypted and protected .omg files can't be possible on another computer. Another solution is to install a directshow filter for WMP, but the problem remain. Before playback, the plug/filter check the DRM rights.

  • ErikS
  • [*][*][*][*][*]
Pre-Test discussion
Reply #62
Quote
Short reply :

- Sorry, I haven't test BeautySlept with and without PNS encoding with musepack.
- Bachpsichord sample is available here :
http://membres.lycos.fr/guruboolez/AUDIO/samples/

I did a quick test, and bachpsichord has the same problem with but not as clear.

ABX results
Beauty slept: 12/12, would give it a score somewhere around 3.5 on the 5 point scale.

Bachpsichord: 10/13, score around 4.0.

  • guruboolez
  • [*][*][*][*][*]
  • Members (Donating)
Pre-Test discussion
Reply #63
I compared the sample Beautyslept at radio profile, and the consequences of the PNS tool (mpc 1.14). I couldn't differenciate the two encodings.

Code: [Select]
1R = C:\Temp\Pns\Sample03\BeautySlept_Q4xlev.wav
2R = C:\Temp\Pns\Sample03\BeautySlept_Q4xlevNOpns.wav

ABX Results:
Original vs C:\Temp\Pns\Sample03\BeautySlept_Q4xlev.wav
   8 out of 8, pval = 0.004
Original vs C:\Temp\Pns\Sample03\BeautySlept_Q4xlevNOpns.wav
   8 out of 8, pval = 0.004
C:\Temp\Pns\Sample03\BeautySlept_Q4xlev.wav vs C:\Temp\Pns\Sample03\BeautySlept_Q4xlevNOpns.wav
   5 out of 12, pval = 0.806
[CODE]


Tried with Bachpischord :

[CODE]
1R File: C:\Temp\Pns\Sample03\Bachpsichord_Q4xlev.wav
1R Rating: 2.0
1R Comment: very grainy
---------------------------------------
2L File: C:\Temp\Pns\Sample03\Bachpsichord_Q4xlevNOpns.wav
2L Rating: 3.0
2L Comment:
---------------------------------------
ABX Results:
C:\Temp\Pns\Sample03\Bachpsichord_Q4xlev.wav vs C:\Temp\Pns\Sample03\Bachpsichord_Q4xlevNOpns.wav
   14 out of 20, pval = 0.058


PNS encodings is worst in this case : sound is grainy, and more unpleasant. I didn't noticed it with BeautySlept. But I must say that I had some difficulties to hear difference between original and -radio (notation : 4.4 and 4.6), less pronounced to my ears than with Bachpsichord (2 and 3/5).

  • Tripwire
  • [*][*][*]
Pre-Test discussion
Reply #64
The reason why I'm asking to include WMA9Pro is that I expect MS to drop WMA9Std in a not so far future in favor of their apparently better Pro version of the codec, for sound quality reasons and also because the backwards compatibility of WMA9Std hindering any further tuning. At least in Q<=25 VBR modes it produces 48-80kbit results (depending on input) that sound pretty OK. I wouldn't wonder if in a year or two, thanks to growing DSP performance in portables, that the Windows Media logo requirements will ask for two channel WMA9Pro support.

Or maybe not. Who knows.

  • amano
  • [*][*][*][*]
Pre-Test discussion
Reply #65
hmm. maybe it sounds like a complete newbie question.

but...
how do I produce a wma 9 pro file????
there is no option in wmp9. or is "windows media audio (variable bitrate)" the one I am seeking?

  • karl_lillevold
  • [*][*]
  • Developer (Donating)
Pre-Test discussion
Reply #66
Quote
We need to define if, for our purposes, Atrac3 = RA or = Sony's version. That matters in the aspect that only one of these is usable in movies and playable in several environments, and only one is playable on hardware players. If both encoders are using the same encoding libraries, we can go ahead. But if they aren't, we'll have to chose between the multipurpose one of the hardware-playable one.

I'll mail Karl asking for clarification.

Hello everybody, so nice to see many familiar names from over on Doom9.

I enjoyed participating in the AAC test and reading the lively discussion, and now I am looking forward to the next test. I think it it is a good idea to include ATRAC3. In many listeners' opinion it does sound better than MP3 at 128 kbps.

RealAudio ATRAC3 is the same as Sony's reference ATRAC3. However, we do not know if this has been updated by Sony since our licensing of ATRAC3. The reason there may be differences is that the reference code does not govern all the settings at various bitrates, which I have been told, may affect for instance the cut-off frequency. My suggestion is to use the ATRAC3 codec that most conveniently enables test participants to create their WAV samples, which would probably be the ACM codec, but I do believe the ATRAC3 flavors are close enough that any choice you are most comfortable with will work out well.

Most of my RV9-EHQ video encodes use RA8 cook at 96 kbps, and I my knowledge and experience in the audio space could have been better.. I did recently put on some high quality head phones and it was nice to hear that my ears could still detect which clips were coded at 128 kbps with AAC  So don't get too upset if anything I say here is not correct. If you need more information I will try to help out.
  • Last Edit: 18 July, 2003, 01:34:14 AM by karl_lillevold
Sr. Codec Engineer (video) | RealNetworks Codec Group | helixcommunity.org 
This information is provided "AS IS" with no warranties,  grants no rights, and reflects my personal opinion.

  • ff123
  • [*][*][*][*][*]
  • Developer (Donating)
Pre-Test discussion
Reply #67
Quote
hmm. maybe it sounds like a complete newbie question.

but...
how do I produce a wma 9 pro file????
there is no option in wmp9. or is "windows media audio (variable bitrate)" the one I am seeking?

Using Windows Media Encoder:

New Session -> Convert a File -> Choose input and output files -> File Archive -> Choose an Audio quality setting (which you will change later)

Go to the Properties tab -> Compression tab -> Edit

Choose WMA Pro as the codec -> Choose bitrate mode (bitrate VBR is the two-pass vbr mode)

Go to the tab next to "General" (mine says 777, standing for the bitrate)

Choose 128 kbps, 44 kHz, 2 channel 24 bit VBR.  The tab will change to 135 (128 kbit/s  audio bitrate plus 7 kbit/s script overhead)

Back in the Compression tab, hit "Apply"

Hit the "Start Encoding" button at the top

This is not very user friendly, is it?  But it seems to be the only way to encode two-pass vbr files, since dbpoweramp doesn't appear to offer this mode.  dBpoweramp is the most convenient way to decode, however, since WME doesn't offer this capability.

ff123

  • amano
  • [*][*][*][*]
Pre-Test discussion
Reply #68
ok. thanks. didn't use the wme ever. so I thought, this option would be integrated in wmp 9.

so, now when I know, how to choose the pro format, I vote to use wma standard in the test, because I guess 99% of the wma tracks out there in the wild will be standard ones, ripped and encoded with wma 8 or 9.

the wme approach seems to be only used by sound pro's that wouldn't use the pro codec with other than experiamental reasons.
and as pointed above, @ 128 kbps hardware compatibility should be one of several reasons.
so drop wma 9 pro and use wma std. just my 2 cents.
  • Last Edit: 18 July, 2003, 10:06:12 AM by amano

  • Nic
  • [*][*]
  • Developer
Pre-Test discussion
Reply #69
The OpenMG software does come with a directshow filter for decoding the ATRAC3 omg files it makes. In graphedit you could easily make WAV files for the ATRAC3 encoded files. Although getting the bitrate to be 128kbps might be a problem.

-Nic

  • den
  • [*][*][*][*]
Pre-Test discussion
Reply #70
Quote
The OpenMG software does come with a directshow filter for decoding the ATRAC3 omg files it makes. In graphedit you could easily make WAV files for the ATRAC3 encoded files. Although getting the bitrate to be 128kbps might be a problem.


I didn't realise there was a directshow filter in there somewhere, but I haven't exactly gone looking for it either. 

I know some people are a little sensitive to the 128 kbit vs 132 kbit situation, but there is no everyday use of ATRAC at 128 kbit, so trying to make it so is pointless.

Surely the additional 4 kbit is not that big a deal for this test.

And hey, if ATRAC3 turns out to be the best, those who love the other formats can always use the extra 4 kbits as an excuse... 

Den

  • rjamorim
  • [*][*][*][*][*]
Pre-Test discussion
Reply #71
Quote
Quote
Anyone with thoughts on this idea? (of separating the test in two)

hm, so you would do 2 different tests (and in use both aac, vorbis and lame)? or one big "mass test" and the second one just a small test with some pros?

which codecs would participate in the mass test? aac, vorbis, lame, wma9 (std or pro?) [and realaudio or sony's atrac3 or real's cook  ]

Well, my idea would be to do two separate (and independent)  tests, one to appeal the masses with codecs of mass consumption, and other to appeal HA with codecs mostly used here.

The Masses test would compare QuickTime, Lame, Sony´s Atrac3, WMA STD and Vorbis

The HA test would compare Nero AAC, Lame, Musepack, WMA PRO and Vorbis.


What do you guys think of this idea?

Regards;

Roberto.


(Damn lousy cybercoffee, connection speed is around 2kBps)
Get up-to-date binaries of Lame, AAC, Vorbis and much more at RareWares:
http://www.rarewares.org

  • amano
  • [*][*][*][*]
Pre-Test discussion
Reply #72
sounds great to me. and by having mp3 and vorbis in both tests, we could perhaps draw some conclusions (ok, I know, not really valid ones) between both tests.

I vote, to perform it as suggested, but to begin with the test for the masses.



EDIT:
upps, we forgot our anchor.

new proposal masses: Lame, QuickTime, WMA STD, Vorbis and Blade as anchor

new HA: ATRAC3, Nero AAC, Musepack, WMA PRO and Blade as anchor.
  • Last Edit: 18 July, 2003, 09:42:12 PM by amano

  • den
  • [*][*][*][*]
Pre-Test discussion
Reply #73
Quote
The Masses test would compare QuickTime, Lame, Sony´s Atrac3, WMA STD and Vorbis

The HA test would compare Nero AAC, Lame, Musepack, WMA PRO and Vorbis.


I agree in principle, but why QT in one and Nero in the other?    I must be missing something.

Do we really need an anchor?

Den.

  • bond
  • [*][*][*][*][*]
Pre-Test discussion
Reply #74
Quote
The Masses test would compare QuickTime, Lame, Sony´s Atrac3, WMA STD and Vorbis
The HA test would compare Nero AAC, Lame, Musepack, WMA PRO and Vorbis.

sounds ok,
but i would use quicktime in both tests (it won in a fair match against nero, and nero will get its chance again in the 64kbps test)

or you prepare the two tests in a way so that they can also be compared, for example by using the same test samples, so that it would be like one big test splitted into two parts...


but, i personally would only need a test which includes the best "versions" each codec has to offer (at 128kbps of course):
lame
quicktime
wma pro
I know, that I know nothing (Socrates)