<JohnV> well.. I've gathered here the technical talk in HydrogenAudio regarding speculations whether US 5,214,742 can be avoided in a MDCT codec with window switching: http://www.hydrogenaudio.org/forums/index....showtopic=13531
<xiphmont> JohnV: that patent is not about windows or window switching, it's about the specific method o deciding when to switch.
<JohnV> xiphmont: well, the question is, can it be avoided in a codec which does window switching..
<purple_haese> JohnV: Yes, by using a different method for deciding when to switch.
<JohnV> purple_haese: does Vorbis use "different method"?
<purple_haese> JohnV: xiphmont seems to think so.
<JohnV> seems to think so?
<xiphmont> JohnV: IANAL [I Am Not A Lawyer]
<JohnV> xiphmont: but you base it on what exactly? You are not a lawyer but what is this thinking based on?
<purple_haese> JohnV: In cases you haven't noticed,  the meeting has moved away from the patent discussion. After consultation with the lawyer, we will disclose as much as we possibly can.

Ok, I asked Xiph's opinion about this specific issue discussed in this thread at Xiph's monthly meeting which is open to public. Here was the answer (the whole Xiphmeeting log can be read here):

Quote

Ok, I asked Xiph's opinion about this specific issue discussed in this thread at Xiph's monthly meeting which is open to public. Here was the answer (the whole Xiphmeeting log can be read here):

JohnV, your quote is a bit misleading as it doesn't show that the discussion was already over (after we agreed that we should prepare a statement) for a while and at that point, you were just adding noise to the meeting (for those who haven't read the logs, the comments in the quote were interleaved with the "on topic stuff").

17:24 < xiphmont> Well, I'd have to be the technical side of making such an opinion wit the lawyer.
17:24 < purple_haese> It doesn't have to be elaborate, and it would be a clear sign of good will.
17:24 < derf_> In fact, you're potentially liable for up to triple damages just for reading the thing without consulting your lawyer, for "willful infringement".
17:25 < jack> derf ha sa point here.
17:25 < purple_haese> Yikes.
17:25 < PsyQ> derf: well, how is anyone who is not a lawyer qualified to call something "patent free"?
17:25 < jack> let me ask the lawyer
17:25 < jmspeex> derf_: single/triple damage doesn't change a thing in our case.
17:25 < purple_haese> Well, then the best we can do, unfortunately, is to explain why we can't make specific statements about specific patents.
17:26 < jmspeex> PsyQ: we define "patent-free". It's like free software, RMS decides what qualifies as "free software"
17:26 < jack> purple_haese: that may be the case.  anyway, i'll send an email to tom and we'll figure out where to go from there.
17:26 < xiphmont> yes, we need some sort of statement.
17:27 < PsyQ> jmspeex: free software does not anything to do with "patent free", so I'd leave RMS out of this discussion
17:27 < jack> ok..  stan, you had some changes to libao you wanted to discuss?
17:27 < purple_haese> Businesses that are interested in Vorbis can always take the money they save on licensing and put part of it toward their own patent opinion, or buy some liability insurance
17:28 < PsyQ> ok - so you defined "patent free" as ?  I guess free of your own patents?
17:28 < jmspeex> PsyQ: What I mean is that when the FSF talks about "free software", they also define what they mean exactly by "free software". We should do the same with "patent-free".
17:28 < volsung> Should I jump in now? 
17:28 -!- mackstann [~death@12-217-123-219.client.mchsi.com] has joined #xiphmeet
17:28 < JohnV> well.. I've gathered here the technical talk in HydrogenAudio regarding speculations whether US 5,214,742 can be avoided in a MDCT codec with window switching: http://www.hydrogenaudio.org/forums/index....showtopic=13531

Discussion already over? In the logs one can see yourself talking about the patent issue with PsyQ (Ivan) in the same minute as I started! And where in the logs it said the patent talk ended? And why did it end so rapidly anyway, before people had really time to ask any questions?
I was just adding noise to the meeting by bringing up this HA thread?? Jeez?? You guys are just incredible...

I said you were adding noise to the meeting because you were the only one left that was talking about that at that point in the meeting (yes, the patent talk ended at the time people started talking about the next point).

I was the only one who was talking about that? Monty's and Haese's comments just magically appeared on the log. I see. I shouldn't have answered to Monty or Haese because other Xiph members pushed the discussion away from the topic at hand.

I'd rather say some other people added "noise", because there was a discussion going on between me, Monty and Haese, and it started by me when everybody was still talking about the patent issue. You can clearly see that in the logs. And again, I was not asking any "statements" about what is meant with "patent free". I was asking for coder opinion, even on the general principle level about the issue covered in this thread (even forgetting Vorbis, and talking totally in general level, but I was made to shut up there by Xiph before I had a chance to say anything further).

Thank you for letting me know that I only added noise, since after all, this is no important issue at all, right, and I'm just a troll who was adding noise there, when there was "much more important issues" to cover. At least I now know your position what comes to me..

This all makes me wonder why I ever wasted my time bringing up Vorbis quality issues when 1.0 was in the makings, or why I bother keeping up Vorbis forums here, or why I bother asking clarifications to an issue, which might in the end add hughly positive impact on Vorbis. Why should a troll like me do anything for you guys.

OK, first my main point was not the noise vs. not noise issue, but just the fact that the way you quoted was misleading. Can you at least acknowledge that?

OK, so you were "made to shut up by Xiph". Everyone knows that Xiph is a large, powerful organization with the power to silence people, right?

I don't want to be boring, but i have to say it again:

US PATENTS ARE NOT WORLD PATENTS !!!    

If Vorbis is illegal in US is an US problem.
Sincerly i don't care; I just wanna remember all of you that HA is mostly composed by European users. 

• US5,321,729 - which appears to ALSO be a patent for Window Switching... I'm not sure since I find reading patent filings is quite painful.
• US5,848,391 - which is completely unrelated to Vorbis, but I was wondering whether Musepack infringed on this (I don't know the details of MPC).

I don't want to be boring, but i have to say it again:

US PATENTS ARE NOT WORLD PATENTS !!!    

If Vorbis is illegal in US is an US problem.
Sincerly i don't care; I just wanna remember all of you that HA is mostly composed by European users.  

Some interesting articles follows:

Recent EFF Legal Cases and Efforts

FFII: Software Patents in Europe

...and we won the 1st battle: Parliament Accepts Petitions Against Software Patents 

(note: would this mean that Xiph should move to EU ?)

JohnV,

It seems to me that Monty's unofficial position on this matter is that patent 5,214,742 is not about Window Switching, but a specific method on when to switch (according to his IRC statement).  To answer the question that you asked on IRC, if a window-switching codec CAN avoid infringment with this patent, then Vorbis is doing so.

The problem is that the wording of that part of the patent is a little vague. I think even you are willing to concede that point. Vorbis doesn't use signal amplitude changes directly to control the window switching.  But since I would dare say that all signals are defined by their amplitude changes, Vorbis is at the very least INDIRECTLY using signal amplitude changes.

So there are two different ways of interpreting this patent, one of them shows Vorbis to be infringing.  Which one is right?  No one is saying officially because no one wants to lose credibility if they're wrong.  I'm betting that Xiph is staying away from it, because they stand to lose no matter how it's cut.

Incidentally, I found two other patents that were interesting:

• US5,321,729 - which appears to ALSO be a patent for Window Switching... I'm not sure since I find reading patent filings is quite painful.
  
  
  
  
• US5,848,391 - which is completely unrelated to Vorbis, but I was wondering whether Musepack infringed on this (I don't know the details of MPC).

It seems to me that Monty's unofficial position on this matter is that patent 5,214,742 is not about Window Switching, but a specific method on when to switch (according to his IRC statement). To answer the question that you asked on IRC, if a window-switching codec CAN avoid infringment with this patent, then Vorbis is doing so.

The problem is that the wording of that part of the patent is a little vague. I think even you are willing to concede that point. Vorbis doesn't use signal amplitude changes directly to control the window switching. But since I would dare say that all signals are defined by their amplitude changes, Vorbis is at the very least INDIRECTLY using signal amplitude changes.

Well, defending Xiph position can be done on two basis:

#1 Attacking claim 1 with argument that vorbis do not code spectrum, but residual

#2 Attacking claim 1 with argument that switching criterion is different from "signal amplitude change"

Personally, I think argument #1 can't be very helpful, because the final result before IMDCT in decoder is actual MDCT signal - which is overlapped and added, so the signal is "coded" no matter how the coding representation is done in the encoder/bitstream (LPC+VQ or direct scalar quantization of the coefficients, like in MP3/AAC) - and the patent does not go into the coding implementation at all.

Argument #2 is more serious - and the validity of it is only subject to determining if Parseval's theorem applies - i.e., is checking envelope level change in threshold bands equal to "signal amplitude change"... but we have another problem - patent does not specify whether "signal amplitude change" is in time or frequency domain, or how should checking/triggering be done at all!

It is a very tricky argument I must admit - because, correct me if I wrong, Vorbis triggers to short block if the amplitude level change between consecutive frequency bands in time (meeting claim #8)  - this is ... well... at least looks like "changing of the signal amplitude"

Quote

#2 Attacking claim 1 with argument that switching criterion is different from "signal amplitude change"

As for #2, that spells a big oops for Xiph. I've got a question though. Is there any documentation on how the encoder chooses between short and long window sizes? I mean, I believe you, but I have to see it myself, because that's strong evidence of infringement, IMHO.

"Vorbis is a hybrid transform domain general purpose audio encoder,
like MPEG in some respects (it is rooted in much of the same basic
theory). For the most part, the similarity comes from the fact that
(like nearly all modern audio compressions), Vorbis codes primarily in
the MDCT domain, using nominal 50% overlap with a y=sin(2PI*sin^2(x))
window (the window is a bit unique). The current Vorbis code and
working spec support switching between two block sizes which must
result in power-of-two sized windowed blocks. From here, the details
diverge." ...

From what I've studied about the architecture of the Vorbis codec over the past week or so, it seems to be using some variant of signal-amplitude changes to determine window size, but the use of "envelope pre-clamping" has to also be considered before determining whether this is ultimately infringing on the patent in question, if it (if I'm reading this right) normalizes the signal before the MDCT step. (Does signal = time/domain envelope?) Also, Monty covers some other points which he believes avoids other MPEG-related patents as well in the linked post.

I'm too uneducated to figure out your statement for #1.

As for #2, that spells a big oops for Xiph. I've got a question though. Is there any documentation on how the encoder chooses between short and long window sizes? I mean, I believe you, but I have to see it myself, because that's strong evidence of infringement, IMHO.

European parliament votes _against_ software patents

This is big news for all users and programmers of XviD in the EU: The european parliament has decided that "inventions" which consist purely of software cannot be protected by patents. This is of course just the way it has always been in europe, but lately there had been much lobby work by "important companies" to have that changed. Until a few weeks ago, it looked like software patents would be introduced in europe within a few month.

The plans would most likely have allowed so-called trivial patents like the famous "one-click-patent" by amazon.com to be possible in europe also. This would have been a serious threat to almost all Open Source development in europe, because there isn't the simplest idea in software development -and in particular in video compression- that isn't under US patent by somebody.
For the experts of you: storing motion vectors relative to a predictor is "protected", MPEG-4's 3dimensional VLCs (run,level,last) aren't free, there even is (or used to be) a patent on YUV as a color space for image compression...

After today's decision, it seems that europe will stay free of such
outgrowth, at least for while...

In you are interested, do have a look at the lots of material at http://www.ffii.org/ or simply search google for "trivial patents".

Sorry to be so boring... 

(Note:  This is from August 1999, but the core functionality of the codec has not changed, has it?)

Quote

e.g. if the complete technological details were publicly available more than 20 years ago, then this technology is patent free. If there was a patent before, it has now expired. If there was a (same, similar or encompassing) patent filed more recently, then it doesn't cover this technology.

The thing is,  which is core of this discussion - is that adaptive window switched MDCT filterbanks didn't exist 20 years ago - they were result of work of H.Malvar, J. Princen, A. Bradley and, finally, B. Edler (dynamic switching)  in late 80's.

Do you have the exact text of this patent?