Moderator comment
Currently the Vorbis patent issue covers these threads:
Vorbis development, status & patent issues PART 1 - NON-technical discussion (This thread)
Vorbis development, status & patent issues PART 2 - Technical discussion
/Moderator comment

To my ears, the compromises Vorbis makes at 64kbps are much more obvious than those of some other codecs. Simple as that.

As I noted, the stereo problems are much more annoying over headphones than they would be over speakers, listening off-centre.


What amazes me is when people say (see Slashdot!) "I encode at vorbis q0 and it sounds perfect". Do you encode stereo content? Do you own a pair of headphones? Can't you hear that all the sounds that used to be over there and over there, are now all lumped together in the centre?

Obviously not. They're probably like my wife: just can't hear stereo. Some people, having determined this fact, decide that they're not qualified to comment on audio quality, and leave it at that. But hey - on Slashdot, everyone is an expert! And Vorbis must be the best.

Cheers,
David.

As you point out with the phrase "to my ears," this is all subjective and dependent on personal preference. None of the codecs are transparent at 64kbps, so they all have differences from the original. It comes down to which difference is more annoying, and that's highly subjective.

When I listen to my q0 Vorbis files by themselves, I don't notice the stereo collapse or the high frequency boost. Real Audio's temporal smearing (Proposed Motto: "We make real hi-hats sound like real maracas.") on the other hand is obvious even when I listen to it by itself. This is essentially the definition of differences versus artifacts. Differences are loss of fidelity. Artifacts are obviously foreign sounds. On 99% of my music collection, Vorbis produces differences, not artifacts.

Don't get me wrong, Vorbis needs to improve, and it will improve. Monty is working on Vorbis 1.1 already. However, keep in mind that the current q0 mode is still essentially the same as it was back in RC3, except for noise normalization added in 1.0 and minor tuning bug fixes in 1.0.1. It was intended to be a respectable first cut. And it's patent-free. Yet it has still managed to come out third best, in a close field of leaders, and beating WMA with its tons of money and development time behind it. It's not perfect, but I'd say it's pretty darn good.

Just my two cents,

Carsten Haese
Ogg Traffic Editor, Xiph.org Foundation

Edit: typo fix

On 99.9% of my music collection, Vorbis produces an instantly audible, very annoying HF boost.

It's a difference, and it sure as hell is an artifact. I cannot say that of MP3Pro or HE-AAC.

Vorbis is simply no longer the top contender in low bitrate audio. It's a respectable second ranker, but sure as hell not the best. That's _objective_ data that this test produced.



...and no improvement has been made ever since.

If Vorbis does not develop further, it will fall even more behind. And as for 1.1, I'll apply what Xiph applied to Matroska/MCF: I'll comment when I see the code, not the claims.

Edit: rewordings

It is claimed to be patent free by Xiph. However there are no single official document online or any other documents of patent searches available.

Fraunhofer guys at IBC2003 claimed that Vorbis is indeed infringing patents, but that no action will be taken yet because it's not bothered at this point, because there's not enough financial gains at stake yet. And this time the FhG guys specified at least one infringement: Vorbis' usage of windowed MDCT filterbanks with adaptive window switching.

This just underlines the point that I was making. The distinction between differences and artifacts is often a subjective matter of personal taste. Our preferences are different, so we will always agree to disagree.



I never said Vorbis was the top contender. Thank you for agreeing with me.



Well, there was a small improvement from 1.0 to 1.0.1, but I know your point is that the HF boost is still the Number One problem.



Monty is well aware of this fact.



Well, the seeming lack of movement is par for the course for Monty. Throughout all the previous releases, improvements happened in fits and starts, with each release being significantly better than the previous one.

Also, let's not forget that Monty spent considerable time on implementing portable player support, which was a necessary step towards greater recognition. So, Vorbis may not the best low-bitrate codec on the planet, but it's the best low-bitrate codec that my Neuros player can handle.

Best regards,

Carsten Haese
Ogg Traffic Editor, Xiph.org Foundation

I must say some things regarding Vorbis, WMA, HE-AAC and such...

From the algorithmic point of view, Vorbis is more effective than WMA (which is a dumb MDCT based codec with window switching and perceptual noise substitution, M/S stereo and vector quantization - i.e. - basic stereophonic codec described in 1992 patent, with addition of noise substitution)

Vorbis, from the other hand - has several more tools, like various lossy stereo coding modes, adaptive huffman codebooks (that won't help that much) and ability to 'peel' the bitstream (never really widely adopted)

HE-AAC has everything that these two codecs have but with addition of powerful TNS algorithm, very effective bitstream syntax, SBR algorithm for very low bit rates etc...

So, WMA sounds very good for the basic set of tools it has - because they tuned encoder well (as good as it can be, actually - not that much - Vorbis can of course be better than current state, but - IMHO - it can never be better than HE-AAC without changing the bitstream and adding new tools.

What the test strongly suggests is that for most people, the arti^Wdifferences that vorbis produces, are more annoying than those of MP3Pro and HE-AAC.


Portable support an advantage for Vorbis? Hah! You can't even get a Neuros in Europe.

The fact that the findings aren't available online doesn't mean they don't exist. However, I have asked one of the people that were involved in the patent search if some evidence can be made available to appease constant skeptics like you.

Update: http://lists.w3.org/Archives/Public/www-sv...03May/0059.html contains some information about this.

Of course, you are free to do your own search. Vorbis is completely open, and Patent documents are available online.


Do you happen to have a patent number to go with that allegation? Words are cheap, and FhG can throw around buzzwords as long as they want, but without specific evidence, this is just hollow FUD that puts them in the same league as SCO with their Linux IP claims.

Also, even if there is a patent that covers a similar technique, that doesn't automatically mean that Vorbis is infringing. There might be prior art, or the patent might not apply due to significant differences between the patented method and Vorbis. We will never know for sure until an infringement suit is brought before a court of law, but as you said, FhG is not interested in doing so. Why? They say it's because they don't have anything to gain. I say it's because they have a lot to lose. It's much more fruitful to make hollow claims and scare away the gullible than to fail proving your claim in court and lose a valuable patent.

Best regards,

Carsten Haese
Ogg Traffic Editor, Xiph.org Foundation

Edit: Added link to more info about the patent searches.

FhG is an research institute and they usually do not sue for infringing of their patents - if someone sues Xiph, that could be either Thomson or Dolby, which are in licensing charge for the patented codecs like MP3, AC3 or AAC.

Nobody is doing that because of the simple fact - it would bring more lose than gain anyway - because suing of non-for-profit projects is just not the way things are being done, and anyway - nobody would benefit from that.

And Vorbis is certainly not a competitor to any of those standards - not at the current state of development and market share, at least.

Good luck, this was asked several times before.

Nobody from Vorbis has ever wanted to back up the patent-free moniker with any real facts, nobody wants to give any guarantees.

Vorbis has always used 'patents' as an argument against MPC, but the reality is that the situation with Vorbis is no different than the one of MPC. If you're going to throw mud, expect to receive some too.

Show something tangible to support the claims, not again hollow things we are supposed to believe, just as we were supposed to believe bitrate peeling would be a major feature advantage of Vorbis, like, 4 years ago already?



Difference is, Linux vendors such as HP are idemnifying their clients from SCO's claims. Is Xiph.org going to do the same?

Great, finally we will get something online about this. I and many people are looking forward to this.
And I don't think it is unreasonable to ask some evidence of something which is usually just taken as granted by many people..

I don't have/know the actual patent number, but the description of the patent is pretty clear, so it shoudn't be too hard to verify.

Hehe. The situation is still the same.

Please let me know when you're interested in bringing this discussion back to the rational level on which it began. Until then, I consider this discussion closed and have nothing more to add.

Sincerely,

Carsten Haese
Ogg Traffic Editor, Xiph.org Foundation

Rational level? Sure, here are the facts:

1) Vorbis keeps touting patent-free-ness as a major feature.

2) Nobody is willing to turn up any verifyable evidence that this is true. (And we asked _several_ times.)

3) Claims are made the the contrary.

4) Vorbis claims that is FUD, but is not willing to give any guarantees about
it's patent-free-ness either.

5) Vorbis dismisses other (better) codecs as having 'patent issues'.

You see my big problem here?

US5357594 (Not from FhG, but one of the Ahead AAC developers was questioning this specifically)

I sure hope you will at least provide some patent search documentations you promised.
Thank you.

Actually, it was a nice discussion on IBC2003 regarding this one - although not related to Vorbis, but it applies to Vorbis case as well to my best knowledge.

I find it extremely hard to avoid this one if you are using adaptive window switched MDCT filterbank in your codec.

This one is analysis/synthesis windowing.

US5214742 is windows switching.

I am also wondering about intensity stereo, used in both Vorbis and Speex.

And it may continue to do so until it's been proven to be infringing.

See the link above for explanations.

See my responses above.

That's because Xiph.org can't afford to indemnify anybody. You get what you pay for.

I am not fit to comment on MPC, since I don't know enough about it, but then again, we're talking about low bitrate performance here, so MPC is out the window. And you do agree that the remainder of the competitors, namely MP3Pro and HE-AAC do have patent issues, don't you?

Thanks for listening.

Actually some of the major ones are not, I remember names but at least one of the major ones said they wouldn't.



FhG are masters of FUD, without bringing lawsuits - that is why Lame is hosted on russian servers, they made all the noises and sent the offical solicitor letters, but no lawsuits have even come about - even for mp3...

FhG are never going to be interested in Ogg Vorbis, for a start (correct me if I am wrong) the holding organization has no serious capital to be handed over in damages, and even then the outcry would (at least from Open Source Community) would fund an effective defence.

I didn't promise anything, but if something tangible is available, I'll let you know.

Those probably do not apply to Vorbis as they all talk about using stereo coding in scalefactor bands. Patent numbers: EP0910927 and EP0910928.

About speex: looks like a CELP codec to me, I don't know if that is patented.

Menno

Here's some other 'rational' thing that I have an axe to grind with:

Vorbis keeps touting itself as supporting features such as bitrate peeling, while still not having any usable utility to do so, and in the meantime downplays things like Matroska/MFC because 'they haven't actually made anything yet'.

You are now implying Vorbis is good because it has spent a lot of time on portable support. If I look at the results (0, as far as it applies to me), I have to seriously wonder whether that time would not have been MUCH better spent on keeping the codec up to date.

It went from being the best codec in some areas, with some issues in other areas, and a lot of promise, to something that's no longer the best in any area, still having serious issues in other areas, and having seriously failed to deliver any of the promises. All of this in space of less than 2 years.

Vorbis went from something that was great because it was just good to something that must be kept alive by its 'supporters' spewing FUD at other things.

I know the reason of some of this degradation was mismanagement and that these issues have been 'fixed', but I have seen no proof or indication of any improvement in the situation lately, on the contrary.

I'm in the situation where I am very seriously wonder whether the time I spent on Vorbis was not just wasted work, and I'd hate _that_ to be true.

Well, they never brought lawsuits because everyone they mailed acomplished to what they demanded.

SoloH is dead. 8Hz-mp3 is dead. Blade and Lame are only (legally) distributable in source form...

If someone didn't acomplish what they wanted, it's pretty much probable that FhG would bring lawsuits.

Crap,

MP3Pro and HE-AAC DO NOT have patent issues, they have clear defined patent list, and well established patent pool - so the company that pays knows what are they paying for. There is also an ISO policy about being fair and reasonable and granting everyone license under equal terms.

Well - they achieved what they wanted, and it is not FUD - because to anyone it is clear that any MP3 encoder / decoder is actually using their patents. Unfortunately, that's the way things go - there are some unavoidable things in MP3.

We already got it. I see more than some vague references and second-guessing what AOL though in there. Note that the original search was actually rather limited, not much guarantee there.



Exactly! NOTHING



MPC is still the best codec out there as low as 128kbps. I wouldn't throw it out of the window so quickly, just like all the other Vorbis supporters like to do because it has 'patent issues'.

Lame hosted on russian servers, well sure, but also on many others..
Umm, check http://mitiok.free.fr/

I see at least: France, Finland, Netherlands, Germany, Poland there alone.
Lame is also hosted in many other places, including US servers (at least Afterdawn), and in dozens of other countries as well..

But this was Off-topic, so no need to continue this..

That is exactly what Vorbis doesn't have. Since Xiph never makes any guarantees or shows any proof, companies interested in using Vorbis can never know if some hidden patent will show up tomorrow and bite them.

Funny thing is, even FhGs stuff is dead - Mp3Enc is no longer available.

Well, the hidden patents will bite you anyway. The thing is, we can't even _know_ if what FhG is spouting is FUD or not. Xiph has so far refused to give any useable details about what they did in their patent search. They don't give any guarantees (quoting former CEO). So what's left exactly? Some vague implications and assumptions?

The situation for Ogg FLAC and Ogg Speex is not any better.

It is available - inside MusicMatch, inside Adobe Audition...

FhG was never really about dealing with end users, they would rather deal with companies that license their technologies, and these companies by their turn deal with the users.

So much, that they created a branch (Opticom) to deal with sales and support for end users.

Which is exactly the reason why no serious big company implements Vorbis - it has legally unclear patent status, no serious backing / idemnifying from a big company (see Linux case).

On the other hand, standards like MP2/MP3/AAC/GSM/CELP/AMR/G.723/G.729... have clear and straightforward licensing models, UN bodies (ITU and ISO) behind the guarantee that patents will be granted on equal-opportunity basis, under fair and reasonable terms - in many cases cheap enough for most companies, etc..

So far, there were no "submarine" patents, except one for MP3 which , I think, has been dealt with (either they went into the patent pool, or they just didn't have the stuff they claimed, I dunno)

Also, there is a great deal of research institutes and expert companies behind them making it harder to believe that they contain something unknown.

Rio have implemented it, their holding company DNNA is big and serious.

Ok, so.. is there any reason to say that Vorbis is NOT infringing this patent, and if so, what?

I think it's actually a funny situation. Vorbis is completely free, but anyone willing to implement it needs to do a very expensive patent search to be relatively safe in case of litigation.

Well, so do several big games.

I know from personal experience this is a bit of a snowball effect though, if they see someone implementing it, they assume it's safe and take the jump as well. I seriously doubt Rio did patent searches, or one of the game developing shops did either.

That doesn't prevent the bomb from dropping just as much as it did with MP3 in 1996.

Patents can bite anybody, whether they already have patents on their technology or not. It is the nature of patent law that you can just sit on your invention until enough people use it, and then sue. The simple fact that Thomson Multimedia tell you what patents they do own doesn't give you any guarantee that they're not using any technology that's patented by somebody else. The only difference is that the risk is somewhat smaller, but it's still there.

At least with Vorbis, the situation is extremely clear. All the code is open, so Thomson would have all the evidence they'd need. They say they don't sue because because Xiph doesn't have any money, but why don't they at least try for a cease-and-desist? Why haven't they sent letters of infringement to Monty, like they sent to the developers of free MP3 encoders? I don't have the answers, but this striking disparity makes me wonder.

Because quite simply, FUD is more effective if it is not adequetly countered.

You can still sue when Vorbis gets too big. (And they might do this, even if they _dont_ have any good grounds to do so, see SCO example)

Just a bit off-topic, but it is a very rare case to see something like this what happened just now:

33 User(s) are reading this topic (11 Guests and 2 Anonymous Users)
20 Members: Atlantis, Dibrom, odnorf, rjamorim, upNorth, Ivan Dimkovic, The Link, todd, Messer, Case, c_haese, zokik, rpop, Phenos, phong, phwip, HMage, spoon, krys, ViPER1313

Wrong - in MP3/AAC case, time proven that there are no submarine patents - and, if there is one, usually it is much more profitable for the patent-holder to make an agreement with the patent pool, and get the overall % of the patent-pool money, which can't be the case for vorbis.



Why would they? You were presented with one Dolby patent that Vorbis has a very small chance to avoid - the fact you are not sued or inform that you should stop your work doesn't mean nothing at all - patent holder could sue/warn on its own behalf and will, there is no rule for that.

This does mean for other companies a lot - you didn't invest a $ in proving that you are in the clear - and you expect everyone to trust you (and what is the back up? couple of developers? come on), or to invest hell load of money to check that - which is exactly the reason why your technology has a proliferation near 0...

With that attitude "come.. and check out by yourself if you want, it's free" you are not going to progress a lot...

I think Vorbis development in general and several issues touched here are important to many people and this is why there is such great interested to this thread and such a high posting rate.

I'm sorry if some of my initial posts were rather harsh, but this is of concern to me and many of the issues have not been adequatly addressed, IMHO.

A long time ago (actually, around two weeks), in a galaxy far far away (actually, in some place called Amsterdam)...

Evil AAC Developer #1: There's too much Vorbis loving around for my liking
Evil AAC Developer #2: I agree. We need to do something about it
EAACD1: What can we do?
EAACD2: I know, let's get some People Who Support Vorbis, and brainwash them to support AAC instead!
EAACD1: Yeah! Great idea! Who should we get?
EAACD2: We'll find them on some audio compression forum I guess.
EAACD1: Oh, that's a good one. Let's go for the big fish. Forum Owner From America.
EAACD2: That's a good one. Let's get Forum Administrator From Finland too. We still need one more
EAACD1: Okay, we need to strike at the heart of Vorbis itself. What about Popular Parttime Vorbis Developer From Belgium?
EAACD2: Yeah! That would really screw Vorbis. So how do we go about doing this?
EAACD1: We lure them over here to attend some product show, and then brainwash them!
EAACD2: How do we lure them here?
EAACD1: Easy. Free lollipops.
EAACD2: Yeah! They will never resist.
EAACD1: Okay, now how should we brainwash them.
EAACD2: With lots of lollipop, booze and crack of course, this is Amsterdam after all.
EAACD1: I think it'll be easier if a female tries to convince them to follow the path of the AAC
EAACD2: Good idea, I have this female friend, they will never resist her charms...
EAACD1: Great! Okay, what shall we do with them once they have become our mindless robots?
EAACD2: I know, it's only 2 weeks until the results of some listening test which AAC is bound to win. We unleash our robots then to spread the word about AAC!
EAACD1: Ohh yeah! Vorbis is so screwed!
EAACD1 and EAACD2's eyes gleam with an evil reddish glow...

[SIZE=1]apologies in advance if i offended anyone. it's a joke, really [I]

They were mints, actually!

Menno

The problem is that analyzing whether a technology infringes on a patent is usually a lengthy process that involves lawyers for both sides arguing before a court of law, interpreting their patent, interpreting the technology, etc. I don't have enough insight into patent law (or Vorbis minutiae for that matter, I'm just a software engineer who pretends to be a journalist from time to time) to make an informed comment about this patent, but I will try to get Monty to comment on it. This, of course, means that he'd have to suspend working on Vorbis 1.1 in order to do so

Right, I hope that you or Monty or someone provides some information regarding this. And it shouldn't take so much time if the alleged already made patent search documents become available, so everybody can read what the search documents say about this.

I just talked to Jack on IRC, and he says, unfortunately, this:



He didn't elaborate on the reasons, but as I said before, it probably comes down to the fact that a patent search is a paid service, and the outcome of the patent search is only available to the people that paid for it. I'll still try to get a "layman's opinion" (as opposed to attorney's opinion, since using the word layman to refer to Monty seems very inappropriate) from Monty about the specific patent number that was mentioned, but I'm afraid that not even that would satisfy the skeptics.

So, given this new information (or lack thereof), what would it take to convince you that Vorbis is patent-free?

Proof! Factual data. HA is all about it. If you come here claiming something and don't provide any kind of hard data, you can't expect the readers to take it for granted.

Right, like Roberto said, the issue is not whether you can convince some individual people or not. The issue is that you advertize vorbis as patent free, yet there's awfully little any documents or anything available which actually support this.
Instead there are patent(s) which seem to apply to Vorbis, but no proofs or any documents have been made available to counter the claims (which have been going on as long as I can remember).

So, the question is, is Xiph's advertizing Vorbis as patent free correct or not. This is probably the most important Vorbis related question there is, since being patent free is the single most important claimed feature of Vorbis.

Here's what pisses me off about Vorbis and Xiph.org. Don't tell us a bunch of shit that you can't back up. Maybe part of the problem was the former CEO, I don't know him or what part he played. I can say his press releases were beyond over-the-top as far as cliches and marketing non-sense is concerned.

We read about Vorbis being great, Vorbis being peelable, Vorbis being patent free, Vorbis 1.1 being even better, and Monty being busy. Well that's nice and all, but all I can find on the Xiph.org web pages are Vorbis 1.0+ which is basically the same thing we got over a year ago.

I know it's free and I shouldn't complain, but please don't tell me things you can't back up. The best way to market your codec is to make it the best. Otherwise get in line.