WMA support

[baobab68]

I was actually snooping around in the Patch Tracker with the intent of finding out what had been done so far, and there I find Marsdaddy has uploaded his WIP effort only yesterday.

Cool.

I'm not a C programmer but I'm going to take a look anyway.

Anyone care to comment about the differences in fixed point versus floating point coding? That's the part I have always wondered about.

[Febs]

You're right, and when I said it I was quoting one of the senior devs of the Rockbox project.

No, you weren't.  

And even if you had quoted someone, what exactly is the point of quoting something from some other thread and then complaining that you are being bashed in this thread by the quote that you yourselve injected into this thread?

[Llorean]

I'm assuming he's referring to the old quote, that I'm paraphrasing here, that says something like "Someone with a clue needs to do the work to convert a WMA codec, and unfortunately most of those with a clue don't use WMA."

This quote happened on a different thread, and I *believe* it happened in different forums (Misticriver) though it is properly attributed to a core developer. But it doesn't call users dumb. There's a difference between cluelessness (which suggests merely lack of knowledge) and dumb (which suggests lack of intelligence) as the former can be fixed.

So what we have is one person's opinion, voiced somewhere else. It is *not* the topic of discussion here, independent of who voiced that opinion.

[Davide-NYC]

Everyone drop it. (please)

I do not use WMA, I never will. But another feather in Rockbox's 'hat' is always welcome.
Rockbox shouldn't be about politics. (or presumed politics) It should just OWN based on it's merits.
It already blows away (in my opinion) all commercial firmwares I've seen thus far, but why rest on one's laurels? We can make it better... together!

WMA decoding would just be one less format Rockbox can't read. Making the switch even more painless for the newbies.  Rockbox users will most likely transition away from WMA in time anyway based on the merits of other formats... but they have to become Rockbox users in the first place. If their whole music library is WMA this will never happen until WMA decoding works in Rockbox.

The last remaining decoder after WMA would be M4A in real time. Then we can simply shake our head at the detractors of Rockbox. (do they even exist?)  

The release of some semi-working code is very welcome news. I know there is a ton of work to do so good luck to everyone and thanks for all of the hard work!

[baobab68]

Davide has a point. Whether WMA is liked or not liked, cool or not cool, it is a stumbling block for a lot of people in converting to ROCKbox.

For some people the ability to go into Windows Media Player and just rip songs is important, without having to know about lameenc --preset insane etc etc

Imagine Apple's lack of dominance if you had to do anything like that in order to successfully use an iPod.

We might end up with a similar situation in the imaging world when Microsoft releases its Windows Media Photo format with Vista. A lot more compression than JPG, but once again not an open standard. I guess photographers with a clue won't use it either.

Please notice how this post is not a complaining one.

Oh and Davide, cool avatar. That scene scared the hell out of me when i was a kid.

[Yotto]

Isn't WMA generally used to enforce DRM?  I could be mistaken, but that was always my impression.  Sure, you can encode in WMA and *not* use DRM, but a lot of WMA music that's sold is sold in WMA specifically becasue it allows DRM.

And rockbox can't legally break DRM, so it can't support WMA with DRM, so even if it supports WMA in general a bunch of people will complain that it doesn't supoprt *their* WMA files.

I don't buy DRMmed music, so I am by far not an expert on this, it's just what I've come to believe.

[saratoga]

Isn't WMA generally used to enforce DRM?  I could be mistaken, but that was always my impression.  Sure, you can encode in WMA and *not* use DRM, but a lot of WMA music that's sold is sold in WMA specifically becasue it allows DRM.

Sure, but the same could be said of any format.  Nothing stops anyone from putting AAC, MP3 and maybe ogg (though I don't know what the license is like for Vorbis) inside an encrypted container.  The choice of fomat is largely irrelevent, since you can encrypt anything you like regardless of format.

[Yotto]

Yeah, but the difference is, with MP3, if you pick a thousand random mp3 files from a big bucket, the probability of picking one that has some form of DRM is pretty low.

Now AAC is a different story, considering it's Apple's preferred file for iTunes, right?  So I take back what I said if we support that

[Llorean]

AAC is actually not an Apple format. It's an audio format from the MPEG-4 standard that Apple chose to use for their own purposes.

[Marsdaddy]

Anyone care to comment about the differences in fixed point versus floating point coding? That's the part I have always wondered about.

Why yes... not quite sure what angle you want covered but...

The existing open source WMA decoder used float heavily - which is fine, unless you are compiling on a platform that doesn't use floats of for which the floating point emulation is slow. I believe most if not all "MP3" players and other mobile devices (such as phones) fall into this category.

So - if you want to port floating point code to a platform that does not use floating points, you have to compromise on accuracy and implement some kind
of "fixed point" scheme where the digits either side of the radix point are held in specific bits of a fixed bit value (i.e. 64 bits in the case of my WMA port).

Take a look at http://en.wikipedia.org/wiki/Fixed-point_arithmetic

[baobab68]

Thanks Marsdaddy, that's exactly what I meant. I always let my mind wander off into, "oh yeah, you'd multiply it by 10,000 or something" but I always kept coming back to "but in the end you'd have to un-multiply it"  :-)

I have had a look at the code, and I know everyone round here always says "if you want something done, go off and do it" but as a first programming effort, you've got to admit, it's pretty daunting.

I've found this whole thread pretty humbling actually. As someone who made the 'mistake' of using WMA in the first place (which is why I take criticism of WMA users kinda personally), I would so love to see WMA be a supported codec, but it's honestly such a *huge* learning curve, I have to admit it's beyond me right now.

I guess I would use this opportunity to say three things:

- if you're a new ROCKbox user, and you want a codec that's pretty successful at low bitrates, I've found that OGG is a good one.

- if you want a really portable codec that's likely to work on tomorrow's player as well as it does on today's, maybe think about MP3 with a decent encoder.

- congrats to Marsdaddy for getting this code as far as he has.

bao

[Marsdaddy]

I have had a look at the code, and I know everyone round here always says "if you want something done, go off and do it" but as a first programming effort, you've got to admit, it's pretty daunting.

I admit! As a first programming effort, it's a no-no. As I've said in the comments accompanying the patch, there aren't many issues with the code - only two that I'm aware of (besides speed).

- congrats to Marsdaddy for getting this code as far as he has.

bao

Thanks, nice to be appreciated. I try not to use WMA myself but if rockbox is going to be an easy switch for most people (not just the wise), I think it's a must.

[Davide-NYC]

(Off Topic)

Isn't the current state of m4a playback at the same state?
It works, and the code is in good shape, but it's just too slow and needs optimizsation.
We need some ASM junkies with some extra time over here!  

(back On Topic) What was the topic?

[bk]

(Off topic continued)

The AAC codec is huge, by far the biggest currently in Rockbox. I did a quick LOC analysis once and IIRC the AAC codec alone constitutes around 20% of the total codebase. Optimization will be quite the job for whoever takes it on.