[00:00:00] Speaker 00:
Our next case for argument is 21-1248, Andrea Electronics Corporation versus Apple.

[00:00:07] Speaker 00:
Mr. Bellinger, please proceed.

[00:00:10] Speaker 03:
Thank you, Your Honor.

[00:00:11] Speaker 03:
May it please the court?

[00:00:14] Speaker 03:
William Bellinger on behalf of Andrea Electronics.

[00:00:17] Speaker 03:
And I want to discuss the legal errors that impact two main issues in the patent office's decision, finding claims 6 to 9 unpackable.

[00:00:29] Speaker 03:
The first regards the motivation to combine the Hirsch and Martin references.

[00:00:37] Speaker 03:
It's clear for the board to have concluded based simply on the fact that Person, Skill, and the Art would have considered Hirsch and Martin together without more to find that an adequate motivation to combine.

[00:00:50] Speaker 03:
That's particularly true here, where the board had earlier issued a first opinion

[00:00:55] Speaker 03:
finding that a person of skill in the art would not have combined those two references because the petitioner's argument was, quote, contrary to the expressed disclosure in one of the two references.

[00:01:07] Speaker 04:
That was tied to the window embodiment of Martin, not the sub-window embodiment, right?

[00:01:15] Speaker 04:
The context of why motivation to combine was rejected.

[00:01:20] Speaker 03:
That was included, yes, John.

[00:01:22] Speaker 04:
So it wasn't about the specific proposed combination here.

[00:01:28] Speaker 03:
Respectfully, I think it was.

[00:01:29] Speaker 03:
So if you read the original decision at appendix 2373, 2374, there's a section summarizing Apple's argument.

[00:01:39] Speaker 03:
There's a section where the board summarizes Patner's argument.

[00:01:43] Speaker 03:
That's on page 2373.

[00:01:44] Speaker 03:
And the board goes on to say on page 2374,

[00:01:50] Speaker 03:
Based on the record developed during trial, patent owners' contentions are persuasive.

[00:01:55] Speaker 03:
Then it goes on in the next paragraph to say, nevertheless, we are not persuaded that one of Skill in the Art would have modified Hershey's system based on the teachings of Martin in the particular manner proposed by Petitioner.

[00:02:06] Speaker 03:
And from there, Your Honor, it's correct.

[00:02:08] Speaker 03:
It goes on to analyze the standards.

[00:02:09] Speaker 04:
And that particular manner was the window version of Martin, not the sub-window version of Martin.

[00:02:15] Speaker 03:
Correct.

[00:02:16] Speaker 03:
Although both were presented in the reply brief, the bulk of the Patent Office's first decision relates to a combination using the first subpoena.

[00:02:26] Speaker 03:
However, the error in the second decision, there's only three paragraphs analyzing motivation combined in the remand decision by the PTF.

[00:02:36] Speaker 03:
The first paragraph suggests that the patent office in the first decision had found a motivation to combine based on this language.

[00:02:43] Speaker 03:
Right.

[00:02:44] Speaker 04:
I just want to clarify.

[00:02:45] Speaker 04:
If your argument is that the current board decision on appeal incorrectly found a motivation to combine based on its view that it had already found a motivation to combine these two references to find these claims obvious based on a sub-window,

[00:03:00] Speaker 04:
understanding of Martin, then we can talk about that.

[00:03:03] Speaker 04:
But I just want to also make sure that you're not trying to argue something further than that, which is the board is barred from finding the motivation to combine

[00:03:13] Speaker 04:
in view of the subwindow and monument of Martin, in view of what it already said in the initial final written decision.

[00:03:19] Speaker 04:
You're not arguing with that, are you?

[00:03:20] Speaker 03:
No, Your Honor.

[00:03:21] Speaker 03:
What we're arguing is that the board in the remand petition had to cite some new evidence from either the reply, something that wasn't considered in the first decision.

[00:03:31] Speaker 03:
And what's improper is to make a contrary finding without any citation to record evidence.

[00:03:38] Speaker 03:
And to the extent there is citation to evidence, and this is with respect to the claim nine argument, the only citation is to the declaration that it previously found was directly contrary to the reference.

[00:03:50] Speaker 04:
Well, let's say we agree with you that

[00:03:53] Speaker 04:
The board overread its initial decision when the board, in the decision on appeal here, believed that that had already found the motivation to combine.

[00:04:04] Speaker 04:
It's sufficient for finding these claims obvious.

[00:04:08] Speaker 04:
And let's also assume for a moment that

[00:04:12] Speaker 04:
Our court's decision did not make such a finding either.

[00:04:18] Speaker 04:
But here, this final written decision in front of us today went further and then also did its own independent analysis on the motivation to combine.

[00:04:30] Speaker 04:
So could you?

[00:04:31] Speaker 04:
Yes, Your Honor.

[00:04:34] Speaker 03:
In that limited context, there's one paragraph, Appendix Page 18 of the board's second decision, and that we would submit is entirely conclusory.

[00:04:44] Speaker 03:
There's no citation to record evidence.

[00:04:46] Speaker 03:
There's no reasoned articulation as to why the board believed.

[00:04:50] Speaker 03:
a skilled artisan would combine Hirsh and Martin, and so under this court's evasive case and other cases, that analysis, that one paragraph analysis with a conclusory statement is legally insufficient.

[00:05:04] Speaker 00:
Well, most of it's conclusory.

[00:05:07] Speaker 00:
In fact, it's only three sentences long, one sentence of which is just a generic sentence from KSR.

[00:05:14] Speaker 00:
But there is one specific thing they say, which is,

[00:05:20] Speaker 00:
there's an express suggestion in Hirsch to look at Martin's teachings.

[00:05:25] Speaker 00:
I guess I imagine there could be cases in my mind where that sentence all by itself would be enough.

[00:05:37] Speaker 00:
Like suppose that in Hirsch the statement wasn't there are disadvantages to

[00:05:43] Speaker 00:
things that look at longer blocks of time.

[00:05:45] Speaker 00:
But suppose instead the statement was, you may also want to include Martin's algorithm in here.

[00:05:55] Speaker 00:
That kind of express reference, if that were what existed in Hirsch, they could probably satisfy their obligation to analyze the motivation to combine by simply saying,

[00:06:07] Speaker 00:
That express reference provides an express motivation to combine, right?

[00:06:11] Speaker 03:
Yes, Your Honor.

[00:06:12] Speaker 03:
And I would agree with that.

[00:06:13] Speaker 00:
So this is cursory because of the facts in this case, which aren't.

[00:06:18] Speaker 00:
But in my mind, there could be circumstances where these three sentences could be enough.

[00:06:24] Speaker 03:
Yes, Your Honor, and I think that's a fair characterization of some of the cases Apple cites in their brief, where you have a base reference and a citation to another reference suggesting incorporation just within the four corners of the reference to say that both of those are part of the same algorithm or approach or the same medical device as in New Vesa.

[00:06:43] Speaker 00:
But here, the only reference in Perch 2 Martin

[00:06:47] Speaker 00:
is one where it talks about disadvantages of systems like Martin's.

[00:06:51] Speaker 02:
Yes, that's exactly.

[00:06:53] Speaker 00:
And so the board needed to do just a little more to say why a skilled artisan would nonetheless take the very thing the reference points to as one of the types of systems for which there are disadvantages and pull it in.

[00:07:08] Speaker 03:
Yes, sir.

[00:07:09] Speaker 03:
That's, I think, a fair summary of our position here.

[00:07:12] Speaker 00:
But they could.

[00:07:13] Speaker 00:
It's just not for us to do.

[00:07:15] Speaker 00:
I noticed in Apple's brief at one point, they said something like, well, the Martin string is only like 0.6 seconds or something.

[00:07:23] Speaker 00:
0.6 seconds seems really long to me, in all honesty, in this kind of art.

[00:07:27] Speaker 00:
So maybe that would be something an expert you could proffer would still say fits into the disadvantages category.

[00:07:35] Speaker 00:
But we don't know.

[00:07:36] Speaker 00:
We don't have that evidence in front of us, do we?

[00:07:40] Speaker 03:
Respectfully, I think that was discussed in the initial petition.

[00:07:46] Speaker 03:
But the way Apple's expert addressed that issue was by assuming a single sublet death, which was what the board previously concluded was.

[00:07:53] Speaker 00:
The first portion.

[00:07:54] Speaker 00:
Correct.

[00:07:55] Speaker 00:
But I guess what I'm saying is the board didn't make any fact findings about whether Martin is only 0.6 seconds and whether that

[00:08:04] Speaker 00:
would cause it to be looked at by a skilled artisan as one of the things that falls into the disadvantaged category or not, right?

[00:08:10] Speaker 00:
That's the kind of fact finding about which we can't possibly do on appeal and which the board has yet to do.

[00:08:19] Speaker 03:
Exactly right, yes.

[00:08:20] Speaker 04:
What if the Hirsch reference said, you know, we're really trying to do noise cancellation here for these signals, and we're going to do that by spectral subtraction.

[00:08:33] Speaker 04:
We take the audio signal and we convert it to the frequency domain, and then we divide up the signal into frequency bins, and then we're going to use some kind of threshold setting algorithm.

[00:08:47] Speaker 04:
Right?

[00:08:50] Speaker 04:
And Martin does a threshold setting algorithm.

[00:08:57] Speaker 04:
That's one way to do it.

[00:08:59] Speaker 04:
But we've come up with our own.

[00:09:02] Speaker 04:
And ours doesn't require as much of a long window of time in order to calculate the thresholds.

[00:09:12] Speaker 04:
But this is the way we're doing it.

[00:09:15] Speaker 04:
Would that be?

[00:09:18] Speaker 04:
Wouldn't that make using Martin's algorithm for spectral subtraction for noise cancellation an obvious thing to do in light of that description of a prior art reference called Hirsch?

[00:09:35] Speaker 03:
Respectfully, no, because the Hirsch algorithm as described is a complete system.

[00:09:41] Speaker 03:
And so what would be required for finding of the motivation to find to get the claimed invention is why would a skilled artist didn't take parts of Hirsch's algorithm, parts of Martin's algorithm?

[00:09:52] Speaker 04:
I'm sorry.

[00:09:53] Speaker 04:
Let me be clear.

[00:09:54] Speaker 04:
My hypothetical is,

[00:09:56] Speaker 04:
Hirsch describes the overall basic framework of doing spectral subtraction, frequency bins and whatnot, and then says, I, Hirsch, am doing my own threshold setting algorithm.

[00:10:12] Speaker 04:
But Martin does one too.

[00:10:14] Speaker 04:
But I'm going to do one differently than Martin's algorithm.

[00:10:18] Speaker 04:
Martin's algorithm you can use in spectral subtraction.

[00:10:22] Speaker 04:
If that's all Hirsch said,

[00:10:26] Speaker 04:
Why wouldn't it be obvious to swap out Hirsch's algorithm for Martin's algorithm?

[00:10:34] Speaker 03:
In that hypothetical, Your Honor, the additional analysis would be required of Martin as to how the specific algorithm Martin would fit into Hirsch and whether it would have either, as you suggest, work consistently with the Hirsch algorithm or denigrate that algorithm.

[00:10:55] Speaker 03:
There's no evidence in the record

[00:10:58] Speaker 03:
on the multiple sub-window theory as to whether that would be preferable or not.

[00:11:03] Speaker 03:
The only analysis presented by Apple is assuming away this monotonically increasing block, which is a single sub-window.

[00:11:11] Speaker 04:
No, I understand.

[00:11:12] Speaker 04:
You're getting back to the facts of your case.

[00:11:14] Speaker 04:
And I appreciate that.

[00:11:16] Speaker 04:
I'm talking about the case I've cooked up in my head.

[00:11:19] Speaker 04:
It's a hypothetical.

[00:11:21] Speaker 03:
And I apologize if I was missing something.

[00:11:23] Speaker 04:
But I guess the point you're trying to make is that perhaps

[00:11:27] Speaker 04:
You're challenging the premise of the easy substitution of the Martin algorithm for whatever algorithm Hirsch happens to have devised for purposes of doing a spectral subtraction.

[00:11:43] Speaker 03:
Yes, Your Honor.

[00:11:44] Speaker 03:
And specifically, in this context, the board already did the analysis of Hirsch combined with Martin in one scenario and found that someone would not have made that combination.

[00:11:54] Speaker 03:
And so I think the proper analysis requires an analysis of both references and whether they could be combined.

[00:12:01] Speaker 00:
You're into your rebuttal time.

[00:12:02] Speaker 00:
Do you want to keep going, or do you want to save some?

[00:12:04] Speaker 03:
If I could reserve it.

[00:12:06] Speaker 03:
But I want to make sure I answer Your Honor's questions.

[00:12:08] Speaker 03:
That's fine with me.

[00:12:09] Speaker 03:
All right.

[00:12:09] Speaker 03:
Thank you, Your Honor.

[00:12:11] Speaker 00:
tell me how to pronounce your name man I'm glad I asked that was not where I was going mr. F

[00:12:31] Speaker 01:
on behalf of Apple.

[00:12:34] Speaker 01:
The board's obviousness determination should be affirmed because substantial evidence supports the board's finding that when Martin's noise floor algorithm is used in Hirsch's system, the combination meets all of the relevant claim limitations.

[00:12:47] Speaker 01:
I think two straightforward and undisputed points refute Andrea's appeal arguments.

[00:12:52] Speaker 01:
First, there is no dispute, as we are here today, that Hirsch refers to Martin as a known alternative approach

[00:13:01] Speaker 01:
to noise estimation for avoiding the problem of speech pause detection.

[00:13:05] Speaker 01:
That's at appendix pages 7 and 18.

[00:13:08] Speaker 01:
There is also no dispute that the board found that a skilled artisan would have considered Hirsch and Martin together.

[00:13:16] Speaker 01:
That's admitted in the reply brief.

[00:13:18] Speaker 04:
Can I get into my concerns right now?

[00:13:22] Speaker 04:
Sure.

[00:13:22] Speaker 04:
Just reading through the record and the briefing, it feels like,

[00:13:28] Speaker 04:
this obviousness question needs to go back on the motivation to combine question.

[00:13:34] Speaker 04:
Because I don't think there was a finding by the board or by this court in the earlier round of this IPR litigation that

[00:13:46] Speaker 04:
explained or justified why one of skill and the art would have made this particular combination of these two references to arrive at these particular claims as recited in these claims, with all of the details of future minimum, current minimum, current magnitude, and all of that in a sub-window context.

[00:14:10] Speaker 04:
And so that first paragraph of the board's motivation decision here, I can't accept.

[00:14:17] Speaker 04:
And then the second paragraph looks awfully conclusory.

[00:14:22] Speaker 04:
It feels a little bit like a half-hearted attempt to do an independent analysis, because it's quite apparent to the board that all the work had already been done and adopted and is off the table.

[00:14:36] Speaker 04:
So why shouldn't this go back?

[00:14:39] Speaker 04:
Because it's just not clear enough to me why we can support something that looks so conclusory right now in terms of the fact finding on why one would combine Martin and Hirsch to arrive at these particular claims.

[00:14:55] Speaker 01:
So I take that in two pieces, Judge Chen.

[00:14:57] Speaker 01:
The first is to break down a little bit

[00:15:01] Speaker 01:
what was going on in the first written decision, which of course is part of the record and what the board, as you see in the first paragraph, clearly thought with the same panel author writing the opinion that it had considered the first time.

[00:15:16] Speaker 01:
So in the first iteration, the way the board framed the issues was that there were two modifications that issue.

[00:15:22] Speaker 01:
One was, would you plug Martin's algorithm into Hersh?

[00:15:28] Speaker 01:
The second was, would you further modify Martin to get to W equals 1?

[00:15:33] Speaker 01:
And that is described as its own separate modification in various places, including the blue brief, the final written decision at 2375, and in the remand papers.

[00:15:45] Speaker 01:
What the board rejected the first time was the second modification.

[00:15:48] Speaker 01:
It said,

[00:15:49] Speaker 01:
Using W equals 1, we think, is contrary to Martin.

[00:15:53] Speaker 01:
The board took no issue with the first one, the plugging in of Martin into Hirsch.

[00:15:59] Speaker 01:
So when the case comes back down, the board says, we already looked at this.

[00:16:05] Speaker 01:
And there are a number of clues in the first final written decision that show why that first modification, simply plugging Martin into Hirsch, was considered and analyzed the first time around.

[00:16:16] Speaker 01:
The board says, at 2373,

[00:16:19] Speaker 01:
Apple reasons, and these are with citations to both the petition and Apple's expert declaration.

[00:16:26] Speaker 01:
Apple reasons that HRSS not only cites Martin, but identifies relevant benefits it provides, those benefits being avoiding speech pause detection.

[00:16:35] Speaker 01:
Also, Apple reasons that it would have been obvious to modify to incorporate that.

[00:16:39] Speaker 00:
Where are you reading from?

[00:16:40] Speaker 00:
I don't see any of this on 2373.

[00:16:42] Speaker 01:
Sorry, is the first full paragraph in the middle there?

[00:16:45] Speaker 00:
Word Apple doesn't appear anywhere.

[00:16:47] Speaker 00:
You're suggesting you're reading the word apple?

[00:16:49] Speaker 00:
The word apple does not appear in any of these.

[00:16:51] Speaker 01:
I apologize.

[00:16:52] Speaker 01:
I was swapping apple for petitioner.

[00:16:54] Speaker 00:
So you're purporting a quote from a page in the appendix by substituting words.

[00:16:59] Speaker 01:
I apologize.

[00:16:59] Speaker 01:
I thought it was easier to use apple instead of petitioner, but I will use petitioner.

[00:17:03] Speaker 01:
It says petitioner contends, for example, with citations to petition.

[00:17:08] Speaker 00:
Yes, but this is just where they're articulating what your contention are.

[00:17:12] Speaker 01:
Right.

[00:17:12] Speaker 01:
And then on the next page, it says, we agree with petitioner that one of skills in the art would have considered Martin's teachings generally when reviewing the teachings of Hirsch.

[00:17:21] Speaker 00:
Yes, generally.

[00:17:22] Speaker 00:
They're not saying these specific ones.

[00:17:24] Speaker 01:
And that's where I think that the two separate modifications are important.

[00:17:27] Speaker 01:
The modification to plug Martin's algorithm into Hirsch is supported by, among other things, the fact that Hirsch, that Martin says it performs well in a non-stationary noise system.

[00:17:39] Speaker 00:
They then go on to say, nevertheless, we're not persuaded one skilled New Yorker would have modified Hirsch's system based on the teachings of Martin in the particular manner proposed by a petitioner.

[00:17:48] Speaker 00:
Right.

[00:17:48] Speaker 00:
Both of those sentences are in such generality

[00:17:55] Speaker 00:
if any, of petitioners things they agree with and clearly disagree with.

[00:18:00] Speaker 01:
I think that respectfully, Your Honor, on the next page you see a reference to, and this is in the middle of the first full paragraph, the proposed modification eliminates Martin's subwindows.

[00:18:12] Speaker 01:
So when it comes back, that modification of Martin to W equals one is gone.

[00:18:18] Speaker 01:
And the question is, take Martin as it is.

[00:18:21] Speaker 01:
And the board, again, the same panel author.

[00:18:23] Speaker 00:
We agree with Petitioner that one skill in the art would have considered Martin's teachings generally when reviewing the teachings of Hirsch.

[00:18:31] Speaker 00:
A general consideration of teachings doesn't satisfy the need to explain why Martin's particular algorithm would have been substituted for Hirsch's algorithm in Hirsch's noise cancellation system.

[00:18:44] Speaker 00:
It's just it's just saying that to me is no more than a statement than these two Documents are both in the same field of analogous arts.

[00:18:52] Speaker 00:
They would have been considered generally

[00:18:54] Speaker 01:
Well, it's much more than that.

[00:18:57] Speaker 01:
It's that there is evidence which was cited on the previous page that they would have been considered generally, and that they are both a known approach to a common problem.

[00:19:05] Speaker 00:
Cited on the prior page in the section where they're just articulating what your arguments are.

[00:19:10] Speaker 00:
They don't cite that evidence in their analysis.

[00:19:12] Speaker 00:
They don't adopt that analysis as true.

[00:19:15] Speaker 00:
They don't say this argument of petitioners is correct.

[00:19:19] Speaker 00:
They go on for two pages about all petitioners' arguments

[00:19:22] Speaker 00:
in detail, detailing each one of your many arguments, they don't indicate that they're adopting those with particularity.

[00:19:31] Speaker 04:
Is it your view that this board decision, there's an inference here that when you read all these pages, even though it has rejected the combination in the context of Martin on a window to window basis, it necessarily did find a motivation to combine these references

[00:19:53] Speaker 04:
in the context of Martin's discussion of sub windows?

[00:19:57] Speaker 04:
I mean, that's the very argument that the board rejected as being actually raised in the IPR.

[00:20:05] Speaker 04:
So you're basically asking us to conclude that it made a fact finding on an argument that it didn't believe was ever raised.

[00:20:14] Speaker 01:
No, I don't think so, Judge Shen.

[00:20:16] Speaker 01:
I think that what the board is saying is that there are two modifications going on here.

[00:20:20] Speaker 01:
And they are agreeing with the evidence that a skilled artisan looking at Hirsch, which says, I, Hirsch, can be combined with spectral subtraction schemes.

[00:20:30] Speaker 01:
Martin is a known approach to a common problem.

[00:20:33] Speaker 01:
would combine the two.

[00:20:36] Speaker 01:
And then you have further evidence, which the board is also citing, that Martin says it performs well in a non-stationary setting.

[00:20:42] Speaker 01:
So you've got the two next to each other.

[00:20:44] Speaker 01:
You've got one referencing the other as a known approach to a common problem.

[00:20:47] Speaker 01:
And respectfully, post-KSR, as this court said in PGS Geophysical, that is a simple, affirmative case for motivation.

[00:20:55] Speaker 01:
What the board rejected was the second complication.

[00:20:57] Speaker 04:
So let me stop you right there.

[00:21:00] Speaker 04:
Let's put this to the side.

[00:21:02] Speaker 04:
You know, you've got other arguments to make but One concern I have is that there were specific arguments raised against motivation I raised by the other side that I don't see the board having ever addressed at any point in time and and so that makes it

[00:21:23] Speaker 04:
a little bit harder to say that we can adopt an understanding of the board decision that invoked KSR of a simple substitution of known elements and just swapping in one element for another element in the primary reference when there are all these very fact-based arguments raised by the other side that have not been addressed by the board.

[00:21:47] Speaker 01:
I think that's where the context the remand is very helpful.

[00:21:51] Speaker 01:
It's very important because it's true The board's decision is relatively short because I think the way that it should be read especially especially coming from the same panel author is that they thought before you said the same panel, but it wasn't the same panel right only to an author you said it this time but early

[00:22:12] Speaker 01:
If I said that, I misspoke.

[00:22:13] Speaker 01:
I meant the same panel author.

[00:22:15] Speaker 01:
And what the board says is, we thought we already considered this.

[00:22:19] Speaker 01:
But then in the context of this remand, this is a very, very narrow dispute as it comes back down.

[00:22:24] Speaker 01:
This court says, go and consider Martin in the multiple subwindow context.

[00:22:29] Speaker 01:
It comes down.

[00:22:29] Speaker 01:
The parties now, previously there were all kinds of references, all kinds of arguments being made.

[00:22:34] Speaker 01:
There's four claims at issue.

[00:22:36] Speaker 01:
And the parties submit two 10-page briefs and two 5-page briefs.

[00:22:40] Speaker 01:
Andrea dedicates six out of 10 pages of its opening remand brief to motivation to combine and these various arguments.

[00:22:49] Speaker 01:
Apple responds at 25, 17 to 20.

[00:22:52] Speaker 01:
The board then comes back and says, we thought we decided this, but to the extent it is still an issue, we think there is a motivation to combine, and there's no teaching away.

[00:23:03] Speaker 01:
And so we think, yes, it would be better if there were two more sentences.

[00:23:07] Speaker 00:
But we think that particularly in the discrete context of this remand, where you have- Do you think it's OK for a board to say we think there's a motivation to combine that just satisfies their obligation under the APA to give reasoned and bases for review?

[00:23:26] Speaker 01:
No, I don't think that alone would be enough.

[00:23:28] Speaker 01:
But I think in this context, you have two separate- And so the only other thing they said

[00:23:33] Speaker 00:
which is in that final paragraph on page 18, is they point to the fact that I pointed out earlier that there's an express suggestion in Hirsch to look to Martin.

[00:23:43] Speaker 00:
Other than that, there is no specificity or particular thing at all articulated in their analysis on motivation and combined.

[00:23:52] Speaker 01:
There's that.

[00:23:53] Speaker 01:
And then in the preceding paragraph, which I agree should not be framed as this court having resolved the issue, but it nevertheless acknowledges what remains undisputed, which is that the reference to Martin is a known approach to avoid a common problem.

[00:24:08] Speaker 00:
Yes, but see, this is exactly the problem.

[00:24:09] Speaker 00:
They don't disagree that it was a known approach.

[00:24:13] Speaker 00:
Right.

[00:24:14] Speaker 00:
What they disagree with is,

[00:24:16] Speaker 00:
Hirsch goes on to say that these kinds of systems have disadvantages and create problems because they take too long of a signal period.

[00:24:25] Speaker 00:
It's a delay.

[00:24:26] Speaker 00:
And that's the problem.

[00:24:28] Speaker 00:
The board didn't address their argument.

[00:24:32] Speaker 00:
I feel like this would be a very different case if they just said, oh, Martin is a known system in Hirsch.

[00:24:40] Speaker 00:
whose algorithm can be used, right?

[00:24:42] Speaker 00:
That would be a very easy case.

[00:24:43] Speaker 00:
Then the board could just say, well, there's an express suggestion to do this.

[00:24:48] Speaker 00:
But the express suggestion actually suggests that the system is a disadvantage.

[00:24:54] Speaker 01:
Well, Chief Judge Moore, two things.

[00:24:56] Speaker 01:
One is that HRSCH does say Martin can be used.

[00:24:58] Speaker 01:
It says our system can be combined with spectral subtraction schemes, of which one is Martin's.

[00:25:04] Speaker 01:
And the second thing is, again, I agree.

[00:25:07] Speaker 00:
Wait, where does it say that?

[00:25:08] Speaker 00:
Where does it say that?

[00:25:09] Speaker 01:
It does say that.

[00:25:10] Speaker 00:
That'd be great.

[00:25:11] Speaker 00:
Show me.

[00:25:12] Speaker 00:
Go to HRSCH.

[00:25:15] Speaker 01:
It is at 456.

[00:25:19] Speaker 01:
So it says at the bottom of the first column,

[00:25:24] Speaker 01:
sort of going over to the top.

[00:25:26] Speaker 01:
And this was evidence that was cited in Apple's petition, which is cited by the board at 2377 in the first final written decision for the fact that these are routinely combined.

[00:25:38] Speaker 01:
It says, the approaches can be combined with well-known spectral subtraction techniques.

[00:25:42] Speaker 00:
I'm sorry.

[00:25:43] Speaker 00:
I'm on page 455.

[00:25:44] Speaker 00:
Is that where you want me?

[00:25:45] Speaker 01:
I'm sorry, 456.

[00:25:47] Speaker 01:
And so it's at the bottom, the sentence that starts at the bottom of the first column and carries over to the top.

[00:25:52] Speaker 01:
It says the approaches can be combined with well-known spectral subtraction techniques.

[00:25:57] Speaker 01:
And it has previously referred to Martin as a known spectral subtraction technique.

[00:26:02] Speaker 01:
And the board at 2377 cited to Apple's petition where it said, citing to expert testimony, that is another indication

[00:26:15] Speaker 00:
that these two things can be put together okay so I'm just gonna be clear about one thing you have to be more careful when you say things to us it doesn't previously refer to Martin as a known spectral

[00:26:28] Speaker 00:
Subtraction technique, it previously refers to Martin as containing an approach known to avoid problems of speech cause detection and to estimate noise characteristics.

[00:26:39] Speaker 00:
Those two things are not necessarily identical.

[00:26:42] Speaker 00:
So it does not previously refer to Martin if it had said Martin is a known spectral subtraction technique and then later said

[00:26:49] Speaker 00:
The approaches can be combined with well-known spectral subtraction techniques.

[00:26:53] Speaker 00:
You'd be on much better footing.

[00:26:55] Speaker 00:
That's what you claim the reference says.

[00:26:56] Speaker 00:
That is not what the reference says.

[00:26:58] Speaker 01:
That's what the evidence in the record that was cited by the board says a skilled artist would read this.

[00:27:03] Speaker 00:
None in this appeal to us.

[00:27:07] Speaker 01:
In the first final written decision.

[00:27:08] Speaker 00:
On a different combination.

[00:27:11] Speaker 01:
Right, exactly.

[00:27:13] Speaker 01:
There were two modifications at issue the first time.

[00:27:15] Speaker 01:
So it comes back, they say we thought we decided this.

[00:27:18] Speaker 01:
We thought this was a straightforward case of modification.

[00:27:21] Speaker 01:
To the extent there's more at issue, this has been presented to us in a very discreet way.

[00:27:26] Speaker 01:
The issues have been

[00:27:28] Speaker 01:
Engaged fully by both sides multiple times now and we disagree We think the issue that was sent back to us was what does Martin teach?

[00:27:36] Speaker 01:
That's what this court said in its decision That's what the board understood its task to be that's where it spent most of its time It said Martin teaches in the multiple sub window context all of the disputed limitations That's no longer even an issue instead.

[00:27:50] Speaker 01:
They're now coming back and saying oh

[00:27:52] Speaker 01:
that really the problem is this issue that the board thought it had decided, re-engaged, and in the context of a narrow set of remand briefs, had everything in front of it and said, we disagree with you.

[00:28:05] Speaker 01:
We think Apple has carried its burden.

[00:28:07] Speaker 01:
Respectfully, we think that under PGS Geophysical and this court's task to reasonably discern what the agency- Can you turn to page 33 of your brief, please?

[00:28:16] Speaker 01:
Sure.

[00:28:20] Speaker 01:
Oh, excuse me, the red brief, did you say?

[00:28:22] Speaker 00:
Yes, page 33 of your brief.

[00:28:24] Speaker 00:
Yep.

[00:28:28] Speaker 00:
In the final paragraph,

[00:28:30] Speaker 00:
You say, the board found you have a quote.

[00:28:32] Speaker 00:
One of skill in the art would have considered Martin's teachings generally, reviewing Hirsch as Martin is specifically referenced in Hirsch.

[00:28:40] Speaker 00:
And you cite page 2373 to 2374.

[00:28:44] Speaker 00:
That quote's not on 2373 at all.

[00:28:47] Speaker 00:
It's only and entirely on 2374.

[00:28:50] Speaker 00:
Is that correct?

[00:28:51] Speaker 01:
That's correct.

[00:28:52] Speaker 01:
That's because we were incorporating the first part of that quote, which is, we agree with petitioner.

[00:28:58] Speaker 00:
So in the context of the- You didn't cite that.

[00:29:02] Speaker 00:
There's nothing here that says that.

[00:29:04] Speaker 00:
You didn't say that.

[00:29:05] Speaker 00:
In your brief, you say, the board found, and you have a quote.

[00:29:08] Speaker 00:
And the quote is entirely on 2374, correct?

[00:29:12] Speaker 01:
The quote is because the first part of that quote, it was intended to say, is we agree with Petitioner.

[00:29:19] Speaker 00:
But that also wouldn't be on 2373.

[00:29:21] Speaker 00:
That would also be contained on 2374.

[00:29:22] Speaker 01:
2373 is where they go through Petitioner's evidence, citing back to the petition for why a skilled artisan would have considered modifying Hirsch's algorithm.

[00:29:33] Speaker 00:
But they don't say we agree with Petitioner's evidence.

[00:29:37] Speaker 00:
They say we agree with petitioner that one of skilled in the art would generally consider this.

[00:29:42] Speaker 01:
Right.

[00:29:43] Speaker 01:
And respectfully, the question is, can you reasonably discern the board's path?

[00:29:46] Speaker 01:
And we think that what the board rejected.

[00:29:48] Speaker 00:
I'm not reviewing this opinion.

[00:29:50] Speaker 00:
This opinion is not what's in front of me.

[00:29:52] Speaker 00:
But let me continue, please.

[00:29:54] Speaker 00:
You then say, in doing so, the board cited and specifically credited Apple's explanations as to why a skilled artisan would, quote, incorporate Martin.

[00:30:05] Speaker 00:
So you go further.

[00:30:07] Speaker 00:
Did the board specifically credit Apple's explanation, and in particular this quote, incorporating Martin's technique for estimating the noise floor of a signal with HRSA's adaptive threshold calculation to improve operation of HRSA's system in a non

[00:30:23] Speaker 00:
stationary noise settings did they specifically credit that particular thing as their reason for why these two things would be combined as you claim they did specifically that is our that is how we think it should be read when it says on 2374

[00:30:42] Speaker 01:
We agree with Petitioner that one skilled in the art would have considered the teachings generally, because it has just said on the preceding page.

[00:30:49] Speaker 00:
It has just said on the preceding page.

[00:30:51] Speaker 01:
Petitioner reasons.

[00:30:52] Speaker 00:
Seven different things that you reasoned.

[00:30:54] Speaker 00:
Seven different things.

[00:30:55] Speaker 00:
If you start on 2372 with Petitioner's contentions, it goes through a lot of different things.

[00:31:02] Speaker 01:
Right.

[00:31:03] Speaker 01:
Two of which are about the motivation to combine the references.

[00:31:05] Speaker 01:
And that's that middle paragraph on 2373.

[00:31:08] Speaker 01:
One is that Hirsch cites Martin and identifies benefits.

[00:31:13] Speaker 01:
And the second is that, and I want to be precise, practitioner reasons that it would have been obvious to modify Hirsch's teachings, quote, to incorporate Martin's technique for estimating the noise floor of a signal within Hirsch's adaptive threshold calculation to improve the operation of Hirsch's system in a non-stationary noise setting.

[00:31:33] Speaker 00:
That's the basis for it.

[00:31:34] Speaker 00:
I would urge you in the future to be more accurate and careful in your briefs.

[00:31:38] Speaker 00:
Do not cite multiple pages for a quote that appears on only one page.

[00:31:43] Speaker 00:
Do not claim the board specifically credited something when it absolutely did not specifically credit something.

[00:31:49] Speaker 00:
It's misleading, and you should be better than that.

[00:31:52] Speaker 01:
I apologize, Chief Judge Moore.

[00:31:53] Speaker 01:
We were not intending to be misleading in any way, and I accept full responsibility for that.

[00:31:58] Speaker 04:
Can we ask a different question?

[00:32:01] Speaker 04:
Let me understand what the proposed combination is.

[00:32:05] Speaker 04:
Is it a full, complete substitution

[00:32:08] Speaker 04:
of the Hirsch algorithm for the Martin algorithm?

[00:32:11] Speaker 04:
Or is it taking only a piece of the Hirsch algorithm out and substituting in the threshold setting algorithm?

[00:32:22] Speaker 01:
That's right.

[00:32:23] Speaker 01:
Which one?

[00:32:23] Speaker 01:
Is it the latter?

[00:32:25] Speaker 01:
It's the latter, yes.

[00:32:26] Speaker 01:
So the Hirsch system is a noise reduction system.

[00:32:29] Speaker 01:
And there are three different algorithms in Hirsch.

[00:32:32] Speaker 01:
The substitution and the combination here

[00:32:35] Speaker 01:
swaps one of those algorithms, the noise detection algorithm, and swaps in Martin's for that.

[00:32:41] Speaker 01:
And there's no dispute that when you do that, all of the recited claim limitations are the same.

[00:32:45] Speaker 04:
My final question is, why didn't you propose Martin as a 102, or maybe Martin in view of Hirsch, simply because Martin has the algorithm, future minimum, current minimum, and all that, and then Martin specifically

[00:33:04] Speaker 04:
describes as one of its applications for its algorithm is spectral subtraction.

[00:33:10] Speaker 04:
And then Hirsch gives you the basic nuts and bolts of what a spectral subtraction is.

[00:33:17] Speaker 04:
Why didn't you do it like that instead of saying, well, let's start with Hirsch, and let's yank out a piece of Hirsch, and let's put in the Martin algorithm there, even though Hirsch may or may not have denigrated the Martin algorithm?

[00:33:31] Speaker 01:
It's because of the way that the claims all sort of successively depend back to claim one.

[00:33:36] Speaker 01:
So Hirsh, it's no longer disputed.

[00:33:39] Speaker 01:
It anticipates a number of those proceeding claims.

[00:33:42] Speaker 01:
And Martin is a more specific type of algorithm.

[00:33:45] Speaker 01:
So that's why we did the plug-in that way.

[00:33:49] Speaker 00:
OK.

[00:33:49] Speaker 00:
OK, thank you, counsel.

[00:33:51] Speaker 00:
Mr. Schellinger, if you have some rebuttal time.

[00:33:54] Speaker 00:
And since we went over with the appointment, if you need extra time, you can have it.

[00:34:00] Speaker ?:
Thank you.

[00:34:00] Speaker 03:
I wanted to respond briefly to a few of counsel's point.

[00:34:04] Speaker 03:
And I think one that was sort of a pervasive point that counsel was making was trying to separate out the motivation to combine analysis from the claim by claim analysis of how the various references

[00:34:17] Speaker 03:
met each of the claims.

[00:34:19] Speaker 03:
And I would just point, Your Honors, to the case cited on page 13 of our reply brief, which is the Metwest-Facco case, which stands for the proposition that those two inquiries, in fact, are common and must be considered together.

[00:34:34] Speaker 03:
And so when counsel suggests that it's proper to first consider motivation to combine two references in the abstract,

[00:34:43] Speaker 03:
and that there's a separate inquiry into how the references match each claim that's contrary to the Med West backhoe case.

[00:34:53] Speaker 03:
The second point, and I think Your Honor noted this, but just so the record is clear, counsel cited to a portion of Hirsch

[00:35:03] Speaker 03:
as purportedly suggesting that Martin was one of the appropriate substitution or an appropriate algorithm that could be substituted into the Hirsch system, citing to appendix page 4456, which is the last page of the Hirsch reference.

[00:35:21] Speaker 03:
If you note the references, the Martin reference at issue is footnote six in the list of references.

[00:35:28] Speaker 03:
The conclusion which counsel read from

[00:35:30] Speaker 03:
Sites at the end of reference five, which is a different paper, where the Martin reference is cited is on appendix page 453, which is the first page of the Hirsch article.

[00:35:41] Speaker 03:
And the complete sentence reads,

[00:35:43] Speaker 03:
Some approaches are known to avoid the problem of speech pause detection and to estimate the noise characteristics just from a past segment of noisy speech.

[00:35:53] Speaker 03:
There's a citation of three articles, including Martin.

[00:35:56] Speaker 03:
And then the paper goes on to say, the disadvantage of most approaches is the need of relatively long past segments of noisy speech.

[00:36:04] Speaker 00:
And unless I'm mistaken, the three

[00:36:10] Speaker 00:
types of algorithms that are known that have these disadvantages, which are the ones disclosed in the Articles 3, 6, and 7, 6 as Martin, none of them are the approach that the conclusion talks about being substituted in.

[00:36:25] Speaker 00:
So the ones that they expressly indicate there are disadvantages of are not the ones that they cite

[00:36:32] Speaker 00:
as approaches that would be good to supplement or substitute in.

[00:36:35] Speaker 03:
Correct, Your Honor.

[00:36:37] Speaker 03:
That's a footnote five, or article reference five, which is not among the list of three, six, and seven on the first page.

[00:36:44] Speaker 03:
So the end of the conclusion in appendix page 456, that conclusion paragraph ends with a citation to article five.

[00:37:00] Speaker 03:
Another thing counsel said, just to make sure it's clear on the record, is respectfully, we do disagree that both of the references, if considered together, have each and every element of the claims.

[00:37:14] Speaker 03:
That is, when we've appealed from the board's consideration of claim nine, we do not believe that the Martin reference shows the periodic limitations of the claim, particularly with respect to the future minimum.

[00:37:28] Speaker 03:
Given that the board didn't clearly articulate how or why the two references were to be combined, we believe that it's a challenge to review that on this record.

[00:37:43] Speaker 04:
We didn't have your argument during the opening argument.

[00:37:46] Speaker 04:
It might not be appropriate to raise another issue in your referral.

[00:37:51] Speaker 03:
I apologize on that, Your Honor.

[00:37:53] Speaker 03:
And in the interest of the court's time unless your honors have additional questions for me.

[00:38:00] Speaker 00:
I Think both counsel this piece is taken