`571-272-7822
`
`
`
`
`
`Paper 36
`Date: March 8, 2024
`
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________
`
`VERANCE CORP.,
`Petitioner,
`
`v.
`
`MZ AUDIO SCIENCES, LLC,
`Patent Owner.
`____________
`
`IPR2022-01544
`Patent 7,289,961 B2
`____________
`
`Record of Oral Hearing
`Held Virtually: January 25, 2024
`____________
`
`Before KARL D. EASTHOM, DAVID C. MCKONE, and
`IFTIKHAR AHMED, Administrative Patent Judges.
`
`
`
`
`
`
`
`
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`
`APPEARANCES:
`
`ON BEHALF OF THE PETITIONER:
`
`
`Babak Tehranchi
`PERKINS COIE, LLP
`11452 El Camino Real,
`Sand Diego, CA 92130-2080
`
`
`ON BEHALF OF THE PATENT OWNER:
`
`
`Sarah E. Spires
`SKIERMONT DERBY, LLP
`1601 Elm Street
`Suite 4400
`Dallas, Texas 75201
`
`
`
`
`
`The above-entitled matter came on for hearing on Thursday, January
`25, 2024, commencing at 10:00 a.m., via video teleconference.
`
`
`
`
`
`2
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`
`
`
`P R O C E E D I N G S
`- - - - -
`JUDGE EASTHOM: Good morning. This is Judge Easthom. Judge
`Ahmed and Judge McKone are on the video hearing with us. This is
`Verance Corp. versus MZ Audio Sciences, LLC, IPR 2022-01544, U.S.
`Patent 7,289,961 B2, and we’ll refer to that as the ’961 patent. We
`appreciate everyone’s willingness to be here. You have the right to be
`heard, so if at any time you have a problem hearing anything or the lines
`drop, please contact the Board personnel that helped you set this up. When
`you’re not speaking, please mute yourself and introduce yourself when you
`begin speaking, and pause before speaking after we ask a question so the
`court reporter creates a good record. Please also briefly pause after you turn
`to another slide or location in the record so that everyone can catch up. We
`have both parties’ demonstratives, and thank you for filing those. They’re
`helpful for us to resolve this issue, although they’re not evidence, as we
`know.
`Petitioner, you have the burden of persuasion. You have 45 minutes.
`Patent Owner also has 45 minutes. You can also each reserve rebuttal time.
`So, with that Petitioner, why don’t you please introduce yourself for the
`record and let me know what time you want to reserve for rebuttal, please?
`MR. TEHRANCHI: Thank you, Judge Eastman. My name is Babak
`Tehranchi. I’m representing Petitioner Verance Corporation. I’d like to
`reserve 30 minutes, and 15 minutes for the rebuttal. So, start with 30
`minutes and then reserve 15.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`
`
`
`3
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`
`JUDGE EASTHOM: Thank you, Mr. Tehranchi. Okay, who do we
`have for Patent Owner?
`MS. SPIRES: This is Sarah Spires from Skiermont Derby. On behalf
`of MZ Audio Sciences, LLC.
`JUDGE EASTHOM: Welcome, Ms. Spires.
`MS. SPIRES: Thank you.
`JUDGE EASTHOM: You may begin, Mr. Tehranchi.
`MR. TEHRANCHI: Thank you, Your Honor, may I share my
`demonstratives? I know you have it, but it may be easier to just go on,
`follow it.
`JUDGE EASTHOM: It’s totally up to you. We have them, and if you
`want to share them, that’s fine.
`MR. TEHRANCHI: Okay, so let me know if you could see them.
`JUDGE EASTHOM: Yes. Great.
`MR. TEHRANCHI: Again, thank you for the opportunity to present
`this case. In our case, we have three grounds. First ground is based on the
`combination of Srinivasan, Cabot, and Kudumakis. The second ground,
`we’re adding Hobson for some dependent claims. And the third ground is
`based on Kudumakis, Tilki, and Cabot. This is on page two of the
`demonstratives.
`On page three, I have reproduced the claim language, claim 1 and
`claim 4. What I like to point out is that I think both parties agree there is no
`controversy that the actual text that is not in the red box is taught by
`Srinivasan, in the Srinivasan combination. And it basically entails getting
`the audio signal, dividing it into time frames and frequency components,
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`
`
`4
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`then taking at least two of those components and altering the face of at least
`one of those components.
`The added limitations that were added during prosecution was about
`selecting the fundamental tone and the overtone. And that is where the
`combination with Cabot and Kudumakis comes into picture. I’d also like to
`note that there are also watermark extraction claims, which more or less are
`similar to the encoding. They’re decoding the watermark from the
`composite signal.
`On page four of the demonstrative, this is just to show that
`Kudumakis basically has these steps. You divide it up into time frames, into
`frequency components, and then it selects two components. It calls them F1
`and F0. And it changes the phase of one component, and it talks about
`changing the component with a lower amplitude; so, it changes the phase of
`that with respect to a reference frequency component.
`On page five of the demonstratives, as you can see, Srinivasan talks
`about selecting a frequency range of 4.8 kHz to 6 kHz. And it says you need
`to do so because that provides better audibility results. But however,
`Srinivasan also recognizes that embedding watermarks in terms of audibility
`is not perfect; and in fact, acknowledges in that column 11, that we’ve
`highlighted that says, yes, if you embed watermarks and you are to change
`the phase of the component to be zero or 180 degrees, it could result in some
`audible artifacts. And it goes on to explain that there is an alternative
`method where it actually uses certain neighborhoods as opposed to changing
`the phase to be zero or 180. So, there is an acknowledgment that there might
`be some audibility issues, and it proposes some methodology to fix that.
`And also, we want to point out that the language in Srinivasan about
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`5
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`selecting the 4.8 to 6 kHz, we don’t believe it’s an absolute language. In
`other words, going back to figure 2, which is on page four of the
`demonstrative, there is no frequency band requirement in that flowchart, and
`neither is in the claims of Srinivasan.
`Why is that important? That’s because in our proposed combination,
`we may have to move away from that 4.8 kHz to 6 kHz. And in our
`combination, the proposal is based on Cabot, which is basically, they tested
`a setup where they changed the phase of third harmonic with respect to the
`fundamental. They were able to show that these phase changes, if you make
`it to the fundamental -- to the harmonics, to the harmonic series, they do not
`produce audible artifacts for certain phase changes. So, in this experiment,
`they showed that if you change the phase under 22 degrees or up to 15
`degrees, it was absolutely impossible to hear that phase change.
`However, if you go a little bit higher, there might be artifacts. But in
`the conclusion, Cabot says, these are really subtle. And it’s also notable that
`the way Cabot conducted the experiment, it was designed to be able to detect
`the audible artifacts. And what I mean by that is they had pristine laboratory
`conditions where they had listeners that at the beginning, they refused to
`believe that there were any changes to these signals until they actually
`showed them on the oscilloscope and showed the traces. They went through
`this training process where they were able to switch between, basically the
`original content that was not changed and then switch back to the watermark
`content. Not watermark, but the content with a base change. They went
`back and forth as many times as they wanted. And at the end, when they got
`to 22 and a half degrees, after all these experiments, about 60 percent of the
`time they were able to tell that there was an artifact.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`6
`
`
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`
`JUDGE MCKONE: This is Judge McKone. Can I ask you a question
`
`here?
`
`MR. TEHRANCHI: Absolutely.
`JUDGE MCKONE: So, if we accept your argument here that Cabot
`shows a phase shift for the third harmonic versus the fundamental is either
`subtle or inaudible. Are you arguing that Cabot suggests using the
`fundamental and the third harmonic to encode data? Or are you simply
`arguing that Cabot shows that that wouldn’t cause a problem in Srinivasan?
`MR. TEHRANCHI: We’re suggesting both, and specifically the fact
`that a person of ordinary skill in the arts, reading Cabot, would understand
`that if you affected the phase change that Srinivasan is talking about, if you,
`instead of using random frequency components that Srinivasan describes, if
`you were to use the harmonic series and specifically change the third
`harmonics phase, then that would not be audible. It would not produce any
`audible artifacts. And in fact --
`JUDGE MCKONE: But Cabot doesn’t discuss encoding data. Right?
`It’s simply an experiment to show how audible a phase shift would be,
`correct?
`MR. TEHRANCHI: Correct. That’s correct.
`JUDGE MCKONE: Okay. So, do you have any affirmative evidence
`that a skilled artisan would have chosen the fundamental and the third
`harmonic as the candidates for encoding data?
`MR. TEHRANCHI: So, I answer that in two steps. So, one is -- and I
`move to my slide, oh geez, to the Kudumakis slide, which is on page 10. So,
`Kudumakis was about watermarking. And Kudumakis talked about
`selecting the locations of fundamental and overtones for embedding
`
`
`
`7
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`watermarks. So, selecting those locations for watermarking purposes was
`known. Now --
`JUDGE MCKONE: I guess, as an initial matter, Kudumakis is the
`reference that you’re citing to show that you would encode in the
`fundamental and the third harmonic?
`MR. TEHRANCHI: It would use those positions to embed the
`watermarks. And it also stated that it’s also advantageous to use those
`positions because they vary across the content in sort of an unknown
`fashion, randomly. So, it would be harder for somebody who’s trying to
`attack and remove the watermarks to be able to locate them.
`JUDGE MCKONE: Okay. And I just want to confirm that it’s
`Kudumakis and not Cabot that you’re citing to and showing the affirmative
`evidence that the fundamental and the third harmonic are your choices for
`encoding?
`MR. TEHRANCHI: That’s correct. In addition, and again, it’s with
`combination with Srinivasan and Cabot, because Srinivasan says, pick two
`frequencies, any two frequencies, and it shows two different ways of picking
`those. And it’s more or less a random way of picking those. And Cabot
`says, well, it’s really, if you change the phase of harmonic series and the
`third overtone, it wouldn’t be audible or detectable, or it would be very
`subtle. And --
`JUDGE MCKONE: That would rebut an argument that the
`fundamental and the third harmonic would cause audible problem that
`Srinivasan wants to avoid?
`MR. TEHRANCHI: That’s correct. That’s correct.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`
`
`8
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`
`JUDGE AHMED: Mr. Tehranchi, speaking of Kudumakis, you
`talked about randomness. Doesn’t Kudumakis teach not just using those two
`frequencies, but also adding a random offset to those frequencies? Why then
`would a person of skill ignore that teaching and use exactly those
`frequencies?
`MR. TEHRANCHI: Randomness is an added benefit. And in our
`view, it’s not necessary to use that, because then you also need to then
`generate that random offset; you need to communicate that random offset to
`the detector. So, it does improve the randomness, perhaps, but it also adds
`some overhead to the embedding process. So, it is part of it. It could be part
`of it, but we’re not using that for this combination.
`JUDGE AHMED: Okay.
`JUDGE EASTHOM: And just to make clear, this is Judge Easthom,
`none of these references, or your modification doesn’t involve changing the
`phase of the fundamental. Right? You’re not doing anything to the
`fundamental, is that right?
`MR. TEHRANCHI: That’s absolutely correct.
`JUDGE EASTHOM: And then in Srinivasan, Srinivasan selects the
`two overtones based on the fundamental. Right?
`MR. TEHRANCHI: So, Srinivasan just says, pick two. And I’ll go
`back to its figure 2. It basically says two. Now, there are different ways that
`it --
`
`JUDGE EASTHOM: Well, in Srinivasan, I think in column 8, it says
`they’re going to select the highest frequency, the strongest frequency, Imax.
`Right?
`MR. TEHRANCHI: Right. And then they’re going to find some --
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`9
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`
`JUDGE EASTHOM: Right. They shift from that to select two
`overtones, right? They --
`MR. TEHRANCHI: Correct. So, they use that as one reference, and
`then they use the two --
`JUDGE EASTHOM: Generally, I think the patent says if you pick
`the strongest frequency component, that’s the fundamental, typically.
`Right?
`MR. TEHRANCHI: As the reference, correct.
`JUDGE EASTHOM: As a reference. And Srinivasan does that, and
`then they shift it up to a higher frequency from 4.8. That’s the preferred
`range. So, you’re not trying to change the fundamental. And Srinivasan is
`basing overtones off the fundamental, and so is your combination, the way I
`understand it. Am I understanding that correctly?
`MR. TEHRANCHI: That’s correct. So, we’re not changing the
`fundamental or the strongest frequency in the combination. It’s the
`overtone. And we’re very specific that based on Cabot, at least, we’re
`looking at the third harmonic, which has been proven that it doesn’t produce
`audible artifacts.
`JUDGE EASTHOM: And so, the fundamental is just, I mean, the
`third harmonic is just a special case of an overtone, is that right?
`MR. TEHRANCHI: That’s correct. So, what Cabot started, the
`premise was that it talks about in its background that there were studies done
`and the prior studies have shown that if you change the phase of harmonic
`series, and it’s not specific to the third harmonic, you won’t be able to hear
`that phase change. And then it devised a very specific test involving a third
`harmonic in that example, just to show that that is the case.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`10
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`
`JUDGE EASTHOM: And part of that is because a third harmonic is
`symmetrical with respect to the fundamental, according to what your expert
`relies on, and I think it’s Lipshitz or maybe Cox, I’m not sure.
`MR. TEHRANCHI: That’s correct, because, you know, and I think
`that’s in the slide 9 that Lipshitz actually says that because these are
`symmetrical signals, the harmonics are. So, their effects, their audible
`effects in changing the phase is much less pronounced. So, that is the
`premise. Correct.
`JUDGE EASTHOM: But as my colleague was putting, that doesn’t
`have anything to do with embedding phases -- in Lipshitz, does that deal
`with encoding and watermarking?
`MR. TEHRANCHI: It’s not about watermarking. And we’re using
`all of these, Risset and Lipshitz and Cabot to show that changing the phase
`of harmonics would not be audible. So, you put a person of ordinary skill in
`the art, when they’re presented with a scenario that they have to pick two
`frequency components and change the phase of the frequency components,
`as Srinivasan describes in its own technique, they would understand, reading
`these other references, that it’s been known for a long time, that if you
`change the phase of harmonic series, that would have less audible effects
`overall in the watermarking. And that’s obviously something that is a
`benefit. It’s something that you want to do in a watermarking system.
`JUDGE MCKONE: I want to do a quick follow up here. You’re not
`arguing that Risset and Lipshitz affirmatively suggest using the fundamental
`and the third harmonic to encode data. Rather, you’re arguing that if you
`choose the fundamental and the third harmonic, Risset and Lipshitz show
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`
`
`11
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`that it wouldn’t cause audibility problems that Srinivasan might be
`concerned about. Is that correct?
`MR. TEHRANCHI: That’s correct.
`JUDGE MCKONE: Thank you.
`MR. TEHRANCHI: That’s correct.
`JUDGE AHMED: Mr. Tehranchi, going back to Srinivasan, you
`talked about the flexibility. Srinivasan says 4.8 to 6 kHz range. You said
`that Srinivasan teaches some flexibility there. If we disagree with you that
`there is that flexibility in Srinivasan, does your combination still work?
`MR. TEHRANCHI: The combination works because Srinivasan
`relied, again, the technique that it used, it was based on more or less, I
`characterized it as being selecting the two frequencies randomly. So, when
`you select the frequencies randomly, then there might be the case that, yes,
`you should choose that 4.8 kHz to 6 kHz, because that would be a better
`position in terms of audibility. But in the combination, we’re not proposing
`to use random frequency components for effectuating the phase embedding.
`But we’re saying that a person of ordinary skill in the art would have
`understood that they would switch -- they would select the harmonic series
`and they would change the phase of one of the overtones. And in that case,
`what Cabot and Risset and all these other evidence, and our experts testify
`that the phase would be practically inaudible, even if you lowered that 4.8.
`And in addition to that, and that’s another part of this, is that changing that
`frequency range to a lower frequency range, perhaps, to enable using the
`harmonic series, would provide other benefits. So, shifting it to a lower
`frequency range would allow you to embed for telephony applications,
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`
`
`12
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`which we’ve said, it cuts it off the bandwidth for that is about 4 kHz. So,
`Srinivasan’s system won’t work there.
`Using the harmonic series, as Kudumakis explains, would basically
`allow different areas of the content to be embedded in sort of a random
`fashion, and it moves dynamically across the content, so it makes the attack
`more difficult. And also in malicious attacks, one of the basic attacks on
`audio watermarks is low-pass filtering, and most of the energy of the audio
`content and speech content is at the lower frequencies. So, by lowering that
`frequency band, changing it, you would gain a lot of robustness. You would
`prevent a watermark to be removed based on normal processing, based on
`intentional attacks on it. So, there are a bunch of benefits that are gained by
`shifting that frequency range.
`JUDGE AHMED: And going back to Kudumakis. It’s a different
`encoding technique, right? It’s the notch encoding. Why would a person of
`skill in the art look to that? And it does not do it at the exact frequencies. It
`does it off the frequencies. Why would a person look to that when
`implementing Srinivasan?
`MR. TEHRANCHI: SO, again, yes, you’re correct, Your Honor.
`Kudumakis talks about a specific type of embedding algorithm. So, it says
`you use a notch to remove the frequencies and because you -- and then you
`insert your own tones in those locations. And because of that specific way
`of embedding, then it says you don’t change the stronger frequency
`components, but you go to the neighborhood and change one of those. So, it
`is that specific embedding process that’s done that then Kudumakis uses to
`say that you embed at the edges of those frequencies. Now, in our
`combination, we’re not relying on the embedding process of Kudumakis
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`13
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`because Srinivasan already has this phase-embedding technique,
`quantization phase embedding, that says you pick two frequencies, you
`change the phase of those frequencies. So --
`JUDGE MCKONE: I do have a question on Kudumakis still. So,
`Kudumakis discusses choosing its placement of the encoding to be near, but
`not on the fundamental or a harmonic. And it seems to have a description in
`here that you’re choosing those locations because the stronger fundamentals
`that are harmonics, if you do not alter them, they will mask the changes that
`you make to the frequencies nearby. Is that correct?
`MR. TEHRANCHI: So, if you have a strong fundamental frequency
`and then that, because it is -- any strong signal in general, would mask other
`smaller signals that is around them, right. So, if you are changing the
`fundamental itself, that’s certainly the case. And in that case, they’re
`actually removing that fundamental or removing the area of that
`fundamental and then adding their own signals in there. In our combination,
`at the end of proposal is, you change the third harmonic, which in a
`harmonic series, it is the lower amplitude. So, there might be some masking
`there, but we’re not removing anything. And for that reason, you don’t need
`to go to the edges or to the neighborhoods of that third harmonic.
`JUDGE MCKONE: But doesn’t Kudumakis stand for the principle
`that if you are making modifications to the stronger signals, in his case a
`notch rather than a phase change, but making changes to the stronger
`signals, such as the fundamental and the overtones of the fundamental, those
`could cause undesirable audio or audibility problem. Is that a fair reading of
`Kudumakis?
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`
`
`14
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`
`MR. TEHRANCHI: In case, when you remove those components and
`you try to replace those components with something completely different,
`right. In our combination, we’re not removing those. We’re changing the
`phase of the third harmonic. And phase change is practically inaudible in
`most cases. And in case of harmonic series, it’s even better in terms of
`audibility.
`JUDGE MCKONE: Well, my concern here, though, is we have three
`references in the combination. Srinivasan discusses choosing its encoding
`range to be in a range where it believes, or Srinivasan believes it’s inaudible
`or barely audible. And Kudumakis chooses its encoding to be away from
`the fundamental and the overtones so that it doesn’t introduce audibility
`problems. So, we have two of the references discussing potential audibility
`problems with the encoding in the range that your combination would
`suggest. Why should we not be concerned about two of the references
`perhaps pointing in a direction other than your combination?
`MR. TEHRANCHI: So, in terms of Srinivasan, again, it uses, it
`changes --the way it selects the frequencies is more or less random. So, it’s
`not using the harmonic series to affect the phase change. So, for that reason,
`the audibility issue that it ports that you would need to put it in the higher
`end in that 4.8 to 6 kHz, that’s not there when you select the harmonic
`series. In Kudumakis, the reason that you have audibility issues is that
`because you’re removing, you’re doing cancer surgery as an analogy, you’re
`removing the signals, and then you’re adding something completely foreign
`in its place. So, the notch embedding, which was also acknowledged in,
`actually, Srinivasan, it does produce audible artifacts if you’re actually
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`
`
`15
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`removing the signal. That’s why Srinivasan, at part of it says, don’t do that
`cancer surgery, go just change the phase of those signals.
`JUDGE MCKONE: But it seems to me that what you’re arguing is,
`okay, we’ve got two references, Srinivasan and Kudumakis, that seem to
`suggest that your selected placement of the encoding could cause noise
`issues, but that Srinivasan and Kudumakis were incorrect about that; they
`misunderstood, and using Cabot to show that they misunderstood and that
`those placements would be possible without audibility problems. And what
`I’m looking for now is the affirmative reason why we would have chosen, or
`a skilled artisan would have chosen the placement that your combination
`requires.
`MR. TEHRANCHI: Okay, right. Again, so, in Srinivasan, the reason
`that Srinivasan says -- and we don’t agree that it’s absolutely, it has to be in
`that 4.8 to 6 reason that it picked those is because it did not contemplate
`using the harmonic series. So, it’s not that it says don’t put them in
`harmonic -- in those areas. Just it didn’t contemplate, it didn’t think about
`using the harmonic series. And Cabot says, well, if you change the phase of
`the harmonic series, your audibility issues go away. Basically, that’s our
`position in terms of changing the phase of the signal. And when we get to
`Kudumakis, Kudumakis did not talk about using phase quantization
`embedding in how --
`JUDGE EASTHOM: Counsel, can I interrupt you for a second? I’m
`sorry. You have about two and a half, two minutes left in your main case.
`But doesn’t Kudumakis encode at the lower frequencies that my friend was
`asking you about, my colleague?
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`
`
`16
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`
`MR. TEHRANCHI: I’m sorry, Kudumakis, could you repeat the
`question?
`JUDGE EASTHOM: Does Kudumakis’s third harmonic, that’s
`encoding something at a lower frequency, right?
`MR. TEHRANCHI: Kudumakis is not very specific about which
`frequency range that it uses.
`JUDGE EASTHOM: But the third harmonic, I thought, was, I think
`your friend and people are sort of saying the third harmonic generally falls,
`it could fall anywhere, really, but generally it might fall under 4000. I don’t
`know. If the fundamental’s 1000, the third harmonic will be 3000; if the
`fundamental’s 2000, third harmonic will be 6000. So, the third harmonic
`can vary. But Kudumakis generally says encode at the third harmonic or
`right around it. So, it’s encoding at what parties here seem to say are lower
`frequencies below what Srinivasan teaches. Right?
`MR. TEHRANCHI: Right. So, yes, if we embed it in the third
`harmonic, it could fall any -- pretty much in the upper band. So, we are
`changing the third harmonic, so, in situations that the third harmonic is in
`that 4 kHz or above, that may be the issue -- may be the case, but it could
`also fall below that as well.
`JUDGE MCKONE: All right. So, if we accept your arguments that
`encoding, that Cabot shows that encoding at the first -- or the fundamental
`and the third harmonic would be problem free, vis-a-vis Srinivasan and
`Kudumakis. Have you shown more than a skilled artisan could have made
`the combination, when what we’re looking for, the question we have to
`answer is whether a skilled artisan would have made the combination?
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`
`
`17
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`
`MR. TEHRANCHI: I think that both is the case in terms of the actual
`process of what to do it, Srinivasan already talks about that. And we’ve
`talked about that in our papers. Srinivasan has the frequency selection
`capabilities. Kudumakis talks about FFTs and spectrums, et cetera, to be
`able to locate those harmonics; and changing the phase of the components is
`not an issue. It’s within the realm of person of ordinary skill in the art. And
`Srinivasan talks about that as well.
`And in terms of would they be making that combination? And our
`answer is absolutely yes, because Srinivasan just has changed the phase of
`the quantization phase embedding. It’s just a matter of whether a person
`ordinarily skilled in the art would have thought about using the harmonic
`series to do that. And the answer is yes, because Cabot and Lipshitz and
`these other corroborating references showed that if you made that change, it
`would be inaudible, practically inaudible, in most situations.
`I think I may be out of time. Let me just, if I may, unless there are
`other questions, I just want to touch on the Tilki reference, which would be
`similar in this case. So, Tilki is also talking about phase embedding. So, it
`selects two frequency components. And it selects the phase of one
`frequency component with respect to the frequency component in the
`previous FFT bin. So, that the difference is that it says here is how you do
`the selection; this is how you would select it. It also talks about an example
`that you would embed above 2 kHz. Again, this will be, again, evidence or
`something that a person of ordinary skill in the art would see if you were
`comparing it to Srinivasan that said do it in 4 kHz. Here Tilki said you
`would do it above 2 kHz. So, the frequency range has shifted. And again,
`that’s not a big deal, because when you’re doing phase embedding to begin
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`18
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`with, it’s not very audible. And now, if you use the same combination of
`Cabot and Kudumakis to select the harmonic series to embed, those audible
`issues would completely go away, or practically go away. So, that is the
`basic combination.
`One of the issues that the Patent Owner has brought up is that the
`embedding bitrate may go down if you use the harmonic series instead of the
`preceding bin, or doing that phase modification in the Tilki combination.
`And the answer is, yes, that is correct. But there are other ways to gain that
`loss in data rate, namely use additional harmonics in that series. Just don’t
`use the third harmonic. And also, it’s known in the industry, and a person of
`ordinary skill would have known that there are tradeoffs in all of this. And
`you don’t really need a high data rate channel for watermarks. In fact, for
`copy control watermarks, you only need one bit that says copy or don’t
`copy. In other cases, you may need a few bits to show an ID. But in any
`case, there are other references, evidence that we’ve presented in our papers
`that says you don’t really need a high data rate watermark. And there’s a list
`of improvements that you would have obtained by implementing Tilki.
`Again, the robustness, because again, that selected frequency is not fixed. It
`moves within the content. Tilki’s methodology is very simple. And then it
`also provides security in terms of low-pass filtering and malicious attacks.
`So --
`JUDGE EASTHOM: Counsel, you have about 10 minutes left if you
`stop in 20 seconds, if you want to wrap up.
`MR. TEHRANCHI: I think this is where I’m going to stop. There are
`other slides about the actual implementation. How would the person of
`ordinary skill in the art would implement. These are all in our papers. And I
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`19
`
`
`
`IPR2022-01544
`Patent 7,289,961 B2
`
`will not discuss ground two which is based on Hobson analogous art unless I
`may use some of my rebuttal time if Patent Owner brings that up. But our
`papers fairly well cover that second ground.
`Thank you, Your Honors.
`JUDGE EASTHOM: Thank you, Counsel. I’m sorry, did I cut you
`
`off?
`
`MR. TEHRANCHI: No, I was going to ask you do I have about 10
`minutes for my rebuttal?
`JUDGE EASTHOM: Ten minutes, yes.
`MR. TEHRANCHI: Perfect. Thank you.
`JUDGE EASTHOM: And then, you’re welcome. And we’ll hear
`from your friend Ms. Spires, now Patent Owner.
`MS. SPIRES: Thank you, Your Honor.
`JUDGE EASTHOM: Do you want to reserve rebuttal Ms. Spires?
`MS. SPIRES: Yes, I’d like to reserve 10 minutes please.
`JU