Monty gave a presentation at the second annual Southern California Linux Expo in LA on November 22, 2003:
"The State of Open Source and Multimedia: Technical Asphalt, Political Potholes and Tabloid Gossip."
His presentation is now available as speex encoded files:
Playing time: 54min
http://www.vorbis.com/~vanguardist/monty_scale1.spx (5.88 MB)
http://www.vorbis.com/~vanguardist/monty_scale2.spx (5.38 MB)

The presentation is very interesting, although understandably Xiph's point of view is highly emphasized. Some important points are not covered extensively enough, or are simplified in order to favor Xiph's agenda. But certainly interesting for anybody who wants to hear Xiph's view about the current state of multimedia: Ogg, MPEG4, WindowsMedia etc.

Definetly an interesting speech, even though it doesn't offer any new information about Xiph's projects.

dev0

I will comment this more later, but one thing which in my opinion was interesting was that Monty, when talking about the latest audio coding situation, mainly concentraded on Vorbis, Apple and Microsoft.
If I wouldn't know better I'd probably think after listening Monty, that Apple is almost the only one pushing MPEG-4 AAC (with DRM, which quite a lot appeared in Monty's comments regarding AAC). However there are a lot of companies and instances working on MPEG-4 AAC technology securing even with the "internal" competition that it is getting pretty wide coverage and possibility of options. Funny detail is that Apple's DRM was circumvented in the same day Monty made his presentation.

He didn't cover so much video specificly (although he did in general level when talking about multimedia), but in my opinion he fails to see (or probably in reality doesn't, but at least didn't emphasize it) the wide industry support of both MPEG4 and WindowsMedia (mainly again concentrading only on Apple and some words about divx). In MPEG4's case the acceptance comes from the international industry standard which is openly available and is constantly developed in co-operation of many top hightech companies which in order produce state of the art reference standards which evolve all the time. In WM's case the pusher is of course MS.
Xiph's video technology is not in as good state, Theora being based on old generation codec, which have trouble against MPEG4 (A)SP. I'm also afraid that Monty underestimates WM a bit. In any case, Xiph's new generation video, Tarkin, is still very far away.

Vorbis, speex, flac and their audio streaming products are pretty solid. No doubt about that. But still the development is very slow and even the few years touted bitrate peeling is not yet here except as a technology demo.
Interesting detail was Monty's comment about BBC, and how Microsoft people came and practically forced BBC to take only WM and abandon Vorbis. This is not the version heard from a certain BBC employee as I told here. Still, maybe MS did come and forced BBC..

Needless to say, Monty avoided completely any patent or license issues, because those are "non-existant", until someone sues.

Still Xiph's products and the idea is a great option for casual users and in certain special sectors (for example games or smaller scale streaming services), but in my opinion it has currently too little resources in order to really compete with the big names about industry acceptance except in certain special sectors. Also the future is not necessarely all that bright for Xiph, unless it can establish itself better and in believable way in order to get wider scale industry acceptance, but it doesn't help that there's only 1 full time developer. Of course it also depends on Xiph's own definition what is successful enough development for it and considering the resources, it's highly respectful what they have achieved so far.

These are my quick thoughts at this point. Anyway, I'll continue a bit later..

Hello JohnV,

Thanks for your analysis.

IMHO, Vorbis - the strongest weapon againist the MP3. The development rate is too slow, the code base have no special ability to surpass MP3, thus the industry keep talking "support in future" but almost 1 year from release but still no hardware support (iRiver is the only one I know), worse than the time of SVCD roll out.

Not to mention patent issue.

I watch their moves from the very beginning, I feel the Xiph development team maybe overloaded. Time passed by, few updates, can't get majority support ... let us (interested users like me) feel uneasy and worry ... the new codec like AAC get much more attention and faster development, technically it may surpass Vorbis at a much faster rate, then all the time and $ to develop Vorbis will be wasted or say depreciated. Maybe "FREE" is still a reason to use it but who want to use it? the DVD pirates?

Sorry for bad english

Game developers
Radiostations

They will save a pretty amount of money not having to pay royalties...

There is RIO too.

I don't understand - the BBC uses Real Networks technology throughout their website.

It's been reported in UK technical/broadcasting newsgroups that the BBC ogg streaming tests were the project of one person, who left. Nothing more, nothing less. BBC engineers do read and post to these newsgroups, but I think this post was just hearsay.

Cheers,
David.

I don't want to sound like advocate... actually I think consumers should choose whatever is the best for them...

But I must say that, technically, complete MPEG-4 AAC is superior to Ogg Vorbis in any way - HE AAC is definitely better than Vorbis at low bit rates, and at high bit rates good and tweaked LC AAC implementation should also be better.

Not to mention features like scalability (not implemented in widespread commercial implementations - but existing and quite possible, just like peeling feature in Ogg vorbis, although there are perfectly good scaleable AAC implementations available now), error protection for streaming, frame size of 1024 and 960 samples in order to be integrated in speech-coding environment, "low delay" mode for 2-way high quality communication, etc.. etc...

The amount of research is also tremendous... but let's just wait and see

Hardware support surely isn't limited to iRiver alone. Look here:
http://wiki.xiph.org/VorbisHardware

About AAC: Is there any portable player available that supports gapless AAC playback ?

Hi,

I own a NEX II, when I heard that NEX IA will support Vorbis. I email them and they said "in the future", almost half year passed but still nothing come out. I'm really WORRY about Vorbis.

I love their way but also see they're having some problem. So sorry I'm unable to help.

The Xiph patent/license discussion has been discussed earlier so much already, so imo we shouldn't continue with that subject anymore here in this thread.. It's boring now because there's nothing new that hasn't already been said.

Not having listened the whole presentation yet (only first few minutes) Monty seems to be quite a reasonable guy and not a zealot saying "this rules" and "this sucks".

Of course he's not like that. Main developers generally aren't like that at all, at least seriously. The enthusiastic supporters are often a different thing though..

the beta ogg vorbis ipod firmwares that will never be released are interesting

1. About bigger attention to AAC
There were polls about hardware players, I think you didn't read them

Poll 1
http://www.hydrogenaudio.org/forums/index....showtopic=1194&

17% people wait for Vorbis Ogg
Most people have players that will support Vorbis Ogg in near future

This poll 2 is even better
Poll 2
http://www.hydrogenaudio.org/forums/index.php?showtopic=191&

22% chose Vorbis Ogg
6,8% chose AAC

And all these despite such giants as Nokia and others trying to make AAC widespreaded.

2. Please prove that AAC better than Vorbis Ogg on high bitrates, and low bitrate difference will be not ABXable (for most people) after nosie problem fixed.
IMHO, IMHO, IMHO

P.S. As I understand the company you are employed is much more interested in AAC

LOL! Of course noone was waiting for AAC support. Back then, AAC support on hardware players was already available



haha. Do you really believe you can extrapolate HA poll results to what is really going on worldwide?

A reality check for you: WMA is the second most famous codec, losing only to MP3. Still, it only has 2 votes there.

Besides, that thread is very old. When it was started, vorbis was a big hype. Since then, lots of people got disappointed at Vorbis' slow development pace, considering AAC got lots of developer attention at the same time. AAC got MUCH better while Vorbis stagnated.
It's a pity people can't change their vote on these polls.



"IMHO" is contradictory with "for most people"

Here's the low bitrate proof:
http://audio.ciara.us/test/64test/results.html

"after nosie problem fixed", sure, but given Xiph's development pace, that will happen around 2006, I reckon.

The high bitrate proof is scattered among several threads around here discussing the artifacts at high bitrates (that lead to the creation of GT3, BTW)

Regards;

Roberto.

I wasn't speaking about opinion polls, I was speaking about performance.



Hmm...

http://audio.ciara.us/test/128extension/results.html

AAC - 128 kbps: 4.42
Ogg - 140.1 kbps: 4.28

Difference is 0.14 points on ITU scale in favor of AAC, and the bit rate of Ogg is 12.1 kbps bigger... I think it is pretty clear that to reach the same subjective quality Ogg needs higher bit rate.

140.1 is not high enough? Well, on 320 kbps I guess bunch will have problems ABXing MP3 against OGG/MPC/AAC/whatever...

At very low bit rates, Ogg was tested against LC AAC encoder which didn't use lossy stereo (QT 6) - so it is hard to get real data about the max performance of each, but SBR AAC definitely proved to be better than Ogg at 64 kbps.

That's from the subjective perspective. From the objective one, MPEG-4 offers much bigger amount of algorithms and tools developed in the last decade - like predictions (for speech-like and tonal signals), temporal noise shaping (for minimizing pre-echo artifacts due to MDCT window length), spectral band replication (for very low bit rate coding), low delay mode (with 20 ms algorithmic delay - as speech codec, for 2-way communication) , error protection and concealment, etc.. etc.. etc...


I'm not advocating it - I'm just telling it is technically superior, and that's a fact - this does not automatically mean that Ogg is not good - it is, it is very good - but if you draw up technical specs of MPEG-4 and Ogg... well.. MPEG-4 is definitely much more complete and advanced.



I'm first a researcher and fan of audio coding science - and here I usually don't speek as an employee of my company

Thread split.
Patent discussion moved here:
http://www.hydrogenaudio.org/forums/index....showtopic=15845

I'm still looking forward to it, JohnV.

I was running on several days of sleep dep at the show, like I almost always am at shows. *Before* I got on the plane to LAX I was up 32 hours straight printing marketing materials, got no sleep on the flight (thank you screaming kids), spent the night after the flight looking for a Kinkos that was open along with the rest of the Xiph posse (who were having fun keeping me up even longer), got four hours of sleep, and then dragged myself to my talk. I swear my original outline didn't focus on Xiph. But when I get no sleep, I ramble like crazy because I'm trying like hell to speak in complete sentences, and I keep going back to what I know best. Practically every pause you saw was me trying to remember what I was talking about ;-)

No one else seemed to think so, but I thought I gave a lousy talk.

As for AAC, I freely admit Vorbis was leapfrogged during a year of effort spent entirely on embedded dev. We're off embedded dev and back to the core now and we'll leapfrog AAC again like we did the first time. Then we'll see who has the faster catch-up time ;-) By all means evaluate the codecs on what exists *right now*... but don't forget the lead has swapped before and will swap again. I plan to still be working on Vorbis ten years from now or longer. I have more coding ideas today than when I started Ogg or started Vorbis.

Monty

Good news that Vorbis team is planning quality improvements - I am looking forward to test new Vorbis...

Anyway .. the MPEG crew is working as well

'course you are. How many layers of committee do changes have to go through? ;-) I know, you're talking about improvements within the spec (and so am I), but all of AAC's recent amazing advance has been due to spec revision. I've been itching to get out my revision boots myself...

Seriously, You list all these things that Vorbis doesn't have as reasons why it's not as advanced. We evaluated and discarded most of the techniques you listed as 'unneccessary gimmickry' in the past five years... before they appeared in the AAC kitchen sink. "Perfection is reached not when there's nothing left to add... but rather when there's nothing extraneous left to take away". Our poor little minimalist encoder seems to hold its own very well, and Vorbis II will likely be considerably smaller a spec than I.

Monty

Couldn't agree more, actually there are two huge additions to the spec planned in the next year as well, but that's the requirement of scientific development

Lawyers are fine - no submarines so far

Now, on the other hand - you also must admit that there have been changes within the current limits of the specs - FhG is no longer the best encoder at 128 kbps, for example - and QT and Nero are being developed very rapidly and being better and better with each new version - without spec change. After all, Vorbis lost at 128 kbps tests against ... LC AAC ... audio (QT 6.3) which was standardized in 1997.

More-than-year old 64 kbps test, before HE AAC, was won by Vorbis because QT AAC, which was used in the test, didn't use lossy stereo mode - it is a general concensus among major today's implementators (Dolby, FhG, Ahead) that stereo image shouldn't be traded for better spectral fullness, to say in that way - however this is not related to any other coding system, for example Ogg does much better job with lossy stereo at 64 kbps, of course.




Of course, each codec has its planned field and industry... AAC, within complete multimedia framework (like MPEG-4) must have error robustness and concealment because one of its applications might be error-prone streaming over air. I must say that implementing codec-sensitive "smart" error protection scheme might give much better error correction possibility if the error rate is very big, like with cellular or digital AM transmissions, compared to "dumb" (this is a wrong name, I know) error protections implemented on a hardware link level. The best way is to provide one and standardized way to do that, instead of making tons of non-compatible error-protection schemes and that's why ER AAC exists.

Low Delay also has applications, maybe not so widespread, but it is nice to have ability to integrate your codec in 20ms frame-switching systems, like g.7xx speech codecs etc - you might ask do we really need 48 kHz 64 kbps 2-way communication - but I dunno.. maybe hi-fi teleconferencing would require that.. or some kind of tele-music with one member sitting in CA, other in WA and third in one NJ


Anyway... I think we all have valid points - so.. may research the development continue and best system win