hi Guys!

Ok so MP3 is old, tired, simple, and after years of hard work is still one of the best encoding options thanks to the LAME Project. So my question is.... is WMA Standard limited by it's structure or could it become as could as LAME MP3 if MS invested in evolving it in the right direction?

If the answer is "no" what limits it? Is it structurally weaker, stronger, the same. What are the different mechanisms employed?

Thoughts, opinions, and expert voices are welcome?

I am seriously interested in this question so know "WMA sucks" stuff pls.

Thanks,
Fairy

Hi,

I think it's been mentioned that WMA uses (probably? certainly?) methods such as vector quantization. Also, we know that it works in the frequency domain (big news ). We can also have a look at the work of Mr. Melvar (edit: is it the correct name ? -- can't find in google) in the audio coding field, before he went working on WMA.

However, *very* few people (maybe.. nobody?) outside of Microsoft knows the exact coding method and strategies used by WMA. With such poor information it is very difficult to know the precise possibilities of the format.

Does someone here know a bit more ?

It would be very difficult to comment without pissing anyone off or breaking TOS....but

I'm really fond of WMA as some of you might already know

I now have a 60GB Zen Xtra and will be staying with WMA 128 vbr because
a: I can't be arsed to reencode everything at higher bitrate.
b: WMA 128 vbr suits me just fine.

I DO NOT and WILL NOT buy any WMA music online DRM or otherwise and therefore do not feel that I am harming the cause of MP3 etc. in any way.

The ffMPEG library contains a WMA9 Standard decoder. I got a peak and it looked like a mixture of AAC, Vorbis & VQF with large static huffman codebooks and something similar to Vorbis' floor0 encoding. That's only my impression.

It's really hard to tell without further studying WMA.

IMHO the floor0-like / VQF-spectral-envelope stuff is a bad approach. Now, Vorbis makes use of the floor1 encoding and switched from "low resolution spectral envelope" to "low resolution noise floor" whith the floor curve's semantic which is IMHO an advantage since this curve is pretty smooth (This is more like the AAC approach).


Sebastian

I remember, that Ivan Dimkovic stated some time ago, that the WMA std. codec is pretty simple and only stays fairly competive, because it is pretty good tuned.

Maybe this post is interesting for you. Just some random seach.

By the way, WMP 10 includes a new version of WMA encoder, WMA 9.1.

Just as a sidenode: The WMA standard format has remained the same since WMAv2.

thislink refers to thisdowload link. is this some scipt that has to be compiled (I don't have a real clue about that, sorry)? is there a binary somewhere?

It is part of Ffmpeg.

I'll do you a binary; will take a moment.

Edit: a binary: http://kotisivu.mtv3.fi/ware.purkki/

thank you very much.

thanks guys. This is an interesting indication of what people know. I hope others (especially the developers) will share there knowledge.

We have knowledge of WMAv2 coding, by looking at the ffmpeg decoder. (WMAv2 is the bitstream format of WMA standard)

Based on this, I personnaly think that wma can be a little better than mp3 for mid-high bitrates (because it doesn't features the sfb21 problem) and about 20% better at low bitrates (because it features a kind of PNS and seems to have longer windows).

However, decoding is obviously problematic because of the huge VQ tables.

As Ivan pointed out in another thread, we have to remind that WMAv2 is still a quite "basic" encoding scheme, theorically quite inferior to AAC, as an example.
It even shares a stupid mp3 defficiency: MS switching is done on a whole granule basis.

btw, the "Mr transform" of Microsoft is E.Malvar. He designed the WMAv2 structure.

I think that wma pro might have been designed by JJ. (which is kinda "scary" as he already designed a big part of PAC and AAC)

Would it break MS copyrights or patents to write a public domain WMA encoder? (say supplied as non-compiled source)

Would it be worth the effort? As the main argument for improving it is that it is widespread but not as widespread as MP3 (and we already have the LAME project for that).

For example if a public domain encoder became very good perhaps someone would find a way of converting the output to DRMed and then online music stores could at least produce decent quality WMAs.

I know this is heresy, however I am interested in whether the "idea" has any mileage.

Regards,
Fairy

First, the patents: Ms has probably some patents on some tricks used by wma. Anyway, it is sure that WMA is covered by more patents than just the Ms ones.

Because of this, it is not possible to write a "public domain" encoder. But you can still write an open source (ie with source available) one. This would not infringe Ms copyrights in any way.

Note this beeing called a "WMA encoder" could infringe Ms trademarks.

why people care about WMA is beyond my imagination posibilities:

1) MS made
2) Windows only
3) Closed source
4) No reference encoder/oficial documentation
5) DRM
6) NO better quality (than other encoders)

so why care?

What kind of merits do these guys have?

Mainly because it's there, in fact it's everywhere, so it is important. Personally, I wouldn't mind it half as much if it was everywhere and good (6).

As for point (1) I don't limit my options to only certain publishers/developers. I like most people choose on the basis of what does what we want, the way we want it, for the least grief.

Now point (5) is an interesting one. One of the reasons why WMA and Apple's AAC/Fairplay are important is because they support DRM. DRM is necessary for the development of new and easier ways to access media. Whether this means recordings by the BBC (or licenced by them....) or music from Sony. Without DRM the webstores would disappear and commercial recordings of TV shows and films wont be made widely, easily, and legally available on the internet. If the Community had responded by rewarding artists for the music distributed over peer-to-peer networks then maybe we wouldn't have gotten to the stage of needing WMA. Maybe.. maybe not.

So to conclude, if it can be made to sound better then WMA becomes a lesser evil. I like to improve my environment :-)

Fairy

(the best way of not helping WMA is not using it )

EDIT: the best way of helping your environment is teaching people use other better codecs, and supporting open standards.

EDIT2: WMA is as much everywhere (or even less) than MP3, so this is a "so so" reason to use it. (not talking about you, about people that use it)

Because it's the second most supported format, by far.



Henrique Malvar created much of the theory behind MDCT.

J. Johnston was an AT&T engineer and worked heavily on PAC and AAC.

Both are very, very good at their field (signal processing)

so? unless you are planing on working for MS, whatever you plan to do with it won't be oficial neither supported, and even if you do end up working for MS, it does not means they would implement what you want.

I'm watching this thread because I'm curious too. I don't like MS, but I don't hate them either. And in any case, even if I did that wouldn't be a good enough reason by far to not care about what everyone knows to be an extremely popular format, even despite it's current flaws. Hopefully this thread won't get too of track, what with the many MS haters in the crowd whom I'm sure would just love to post their own MS/WMA bash upon seeing these posts. Fact is, no matter how much you hate it, it's not going to go away for a very long time. Thus the only alternative becomes making it better. And then the logical question becomes, can it be? I think so. FWIW, I'm playing with it today, mainly becuase my new player is coming soon and I was curious how it sounded at lower bitrates (thinking extended battery life here).

The thread started as a theoretic discussion. "What is the potential of WMA?"
Then the thread moved onto the question of "Could a community implementation of a WMA encoder be made?"

You went on to ask why bother? The answer is because it is a widely supported CODEC. I agree that MP3 is more widely supported however as WMA is taking a hold in more areas of music distribution it would be good to see it reach it's full potential for audio fidelity. The idea is not that MS would necessarily implement anything it's that an Open Source project could (in theory).

Fairy

LAWE Ain't a WMA Encoder.

i see the point, but i would rather focus on MP3/AAC/Vorbis

<rant>

This thread has quickly degenerated to another MS/WMA bash one despite the original author requests... sad...

What I see is that very few posts since post #16 added anything useful to this discussion.

What annoys me is that whoever (like myself) is interested in this subject is getting nothing but open source zealotry, MS-and-WMA-are-hellraised-and-everyone-who-uses-it-should-die like posts and another bunch of yadda yadda. I'm really disappointed...

Please don't take it personal... I do not intend to engage in fruitless discussions.

Leaving the building now...

</rant>

Right. It seems some people can't keep their childish hatred towards MS to themselves

IMO MS could make the WMA open source. That would help the format become more popular. You know, help make a comunity. In terms of quality both (MP3/WMA) offers almost the same quality. Even though I would stick MP3.




PS: Come on, people. It's just a tech thread. Let's forget about the MS-Hate bull@#$%.

To make it open source would most likely scare all the investors
of these "legit type" music distribution sites like Napster,buy dot com,etc.
The scare of the DRM being compromised would scare off the investors.
Scare the investor,no more business,no more capitalism and Microsoft is
all about capitalism,not doggin it,who don't wanna make money.And so this
somehow ties into the original question...Capitalism and fear has stifled
the progress of technology,and i'm going to stop now before i take myself
too serious.

If Microsoft put more research & development into WMA, I am sure it would sound better than it is now. I personally have alot of my songs in my library encoded in WMA purely becuase it makes it so much easier to transfer them to my portable player and I don't mind if the quality is lower (subjectively) it beats listening to the radio IMHO.

I hope they do pour more research into it, who knows if MS is serious about this push into the music arena via WMP10 they may have something up their sleeve...