`
`Exhibit E
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 2 of 31
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`__________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`__________________________________________________________________
`
`
`
`META PLATFORMS, INC.,
`
`Petitioner,
`
`v.
`
`JAWBONE INNOVATIONS, LLC,
`
`Patent Owner.
`
`
`Patent No. 8,503,691
`Filing Date: June 13, 2008
`Issue Date: August 6, 2013
`
`Inventor: Gregory C. Burnett
`Title: VIRTUAL MICROPHONE ARRAYS USING DUAL
`OMNIDIRECTIONAL MICROPHONE ARRAY (DOMA)
`
`
`
`
`__________________________________________________________________
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`
`Case No. IPR2024-00349
`__________________________________________________________________
`
`
`
`
`
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 3 of 31
`
`
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`TABLE OF CONTENTS
`
`Page(s)
`
`
`I.
`INTRODUCTION ........................................................................................... 1
`THE ’691 PATENT ......................................................................................... 4
`II.
`III. THE RELEVANT ALLEGED PRIOR ART .................................................. 7
`A.
`Brandstein (Ex. 1003) ........................................................................... 7
`IV. CLAIM CONSTRUCTION ............................................................................ 8
`V.
`LEVEL OF ORDINARY SKILL IN THE ART ............................................. 8
`VI. ARGUMENT ................................................................................................... 9
`A.
`Legal Standard ....................................................................................... 9
`B.
`Ground 1: Brandstein Does Not Render Obvious Claims 1-20 .......... 10
`1.
`Petitioner’s Combination Does Not Disclose or Render
`Obvious “the second linear response to noise [is/being]
`substantially similar to the first linear response to noise,
`and the second linear response to speech [is/being]
`substantially dissimilar to the first linear response to
`speech” as Recited in Independent Claims 1, 23, 27, 28,
`29, and 41 ..................................................................................10
`Petitioner Does Not Identify Any Motivation to Combine
`Embodiments and References Within Brandstein ....................19
`Ground 2: Brandstein and McCowan Do Not Render Any
`Claims Obvious ................................................................................... 23
`VII. CONCLUSION ..............................................................................................24
`
`C.
`
`2.
`
`i
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 4 of 31
`
`
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`TABLE OF AUTHORITIES
`
` Page(s)
`
`Cases
`Apple Inc. v. Samsung Elecs. Co.,
`839 F.3d 1034 (Fed. Cir. 2016) .......................................................................... 10
`Application of Arkley, 455 F.2d. 586 (C.C.P.A. 1972) ............................................ 19
`Belden Inc. v. Berk-Tek LLC,
`805 F.3d 1064 (Fed. Cir. 2015) .......................................................................... 21
`Forest Lab’ys, LLC v. Sigmapharm Lab’ys, LLC,
`918 F.3d 928 (Fed. Cir. 2019) .............................................................................. 9
`Graham v. John Deere Co. of Kansas City,
`383 U.S. 1 (1966) .................................................................................................. 9
`KSR Int’l Co. v. Teleflex Inc.,
`550 U.S. 398 (2007) .............................................................................................. 9
`Meta Platforms, Inc. v. Jawbone Innovations, LLC,
`IPR2023-01130, Paper 10 (P.T.A.B. Dec. 28, 2023) ..................................passim
`Meta Platforms, Inc. v. Jawbone Innovations, LLC,
`IPR2023-01130, Paper 12 (P.T.A.B. Feb. 22, 2024) ............................................ 1
`Net MoneyIN, Inc. v. VeriSign, Inc.,
`545 F.3d 1359 (Fed. Cir. 2008) .......................................................................... 19
`Statutes
`35 U.S.C. § 103 .......................................................................................................... 9
`
`
`
`
`ii
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 5 of 31
`
`
`
`Exhibit No.
`2001
`2002
`2003
`
`
`
`
`
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`
`EXHIBIT LIST
`
`Description of Document
`Declaration of Akbar M. Sayeed, Ph.D.
`Curriculum Vitae of Akbar M. Sayeed, Ph.D.
`Excerpts of MICROPHONE ARRAYS: SIGNAL PROCESSING
`TECHNIQUES AND APPLICATIONS, Ch. 18, Future Directions
`in Microphone Array Processing (Michael Brandstein &
`Darren Ward eds., Springer-Verlag 2001) (“Brandstein”)
`
`iii
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 6 of 31
`
`
`
`I.
`
`
`INTRODUCTION
`The Petition raises arguments that the Board already rejected while denying
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`institution of Meta’s Petition in IPR2023-01130. Meta Platforms, Inc. v. Jawbone
`
`Innovations, LLC, IPR2023-01130, Paper 10 (P.T.A.B. Dec. 28, 2023) (denying
`
`Petition for failure to show that Brandstein discloses virtual microphones with
`
`“substantially similar . . . response[s] to noise.”); see also Meta Platforms, Inc. v.
`
`Jawbone Innovations, LLC, IPR2023-01130, Paper 12 (P.T.A.B. Feb. 22, 2024)
`
`(reaffirming decision, denying request for rehearing). This patent includes the same
`
`limitation, and Meta raises essentially the same arguments regarding Brandstein’s
`
`disclosure. Institution of this Petition should be denied for the same reasons.
`
`In the -01130 proceeding, Meta argued that the “substantially similar . . .
`
`response[s] to noise” was rendered obvious by Brandstein’s Fig. 5.2. The Board
`
`rejected that argument, finding that the overall directivity depicted in Fig. 5.2 does
`
`not disclose the response of any component virtual microphone, and because other
`
`disclosure, including Fig. 5.3, “refutes Petitioner’s assertion [that] the virtual
`
`microphone noise responses are substantially similar.” IPR2023-01130, Paper 10 at
`
`24.
`
`This Petition presents the same deficiencies the Board identified in IPR2023-
`
`01130. The Board already found that Meta’s reliance on Brandstein’s Fig. 5.2 was
`
`misplaced, and that Brandstein’s other discussion, as well as Fig. 5.3, refute Meta’s
`
`1
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 7 of 31
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`arguments that the virtual microphones of Brandstein’s Griffiths-Jim beamformer
`
`(“GJBF”) have substantially similar responses to noise:
`
`We agree with Patent Owner that the record does not
`support Petitioner’s assertion that Brandstein teaches or
`suggests the independent claim requirement, “wherein the
`first virtual microphone and the second virtual microphone
`are distinct virtual directional microphones with
`substantially similar responses to noise.” Contrary to
`Petitioner’s interpretation of Brandstein Figure 5.2, that
`figure does not lead to an inference that the individual
`microphone responses have similar responses to noise.
`Prelim. Resp. 12–13. In effect, Petitioner is arguing that
`knowing that the sum of two numbers is equal to “2”
`necessarily means that each number is “1”, as opposed, for
`example, to one number being “2” and the other “0”.
`
`Moreover, as discussed above, Brandstein’s description of
`Figure 5.1 demonstrates that the virtual microphones of
`Brandstein do not have “substantially similar responses to
`noise.” Prelim. Resp. 9–11. Likewise, Figure 5. 3 of
`Brandstein also refutes Petitioner’s assertion the virtual
`microphone noise responses are substantially similar.
`Prelim. Resp. 13.
`IPR2023-01130, Paper 10 at 24-25. The Board reaffirmed its finding in denying
`
`Meta’s request for rehearing, stating that:
`
`[Figure 5.2] does not lead to an inference that the
`individual microphone responses have similar responses
`to noise. . . . Thus, Petitioner’s sole basis for arguing that
`Brandstein teaches the noise response requirement is
`flawed. . . . Even if we were to agree with the construction
`of the noise response requirement now asserted by
`Petitioner in the Request, Petitioner’s flawed reliance on
`Figure 5.2 of Brandstein still undercuts all of the
`unpatentability challenges raised by Petitioner, inevitably
`
`2
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 8 of 31
`
`
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`leading to the Board’s Decision Denying Institution of
`Inter Partes Review.
`IPR2023-01130, Paper 12 at 5-6. The Board further found that “Brandstein’s
`
`disclosure does not support Petitioner’s assertion that the virtual microphones of
`
`Brandstein have substantially similar responses to noise,” citing to Patent Owner’s
`
`declarant. Id. at 8-9.
`
`
`
`If anything, Meta’s argument in this Petition (Paper 1 (“Petition” or “Pet.”))
`
`is even more lacking. The Petition paraphrases the same arguments in IPR2023-
`
`01130, while hand-waving the requirement that the virtual microphones’ responses
`
`to noise be substantially similar and dedicating the bulk of its analysis to the “speech
`
`response” limitations instead. Pet. at 32-34. But the Petition’s heavy reliance on
`
`Brandstein’s Fig. 5.3 in the context of the speech responses makes Meta’s failure to
`
`address that figure in the context of Brandstein’s noise responses even more
`
`conspicuous. In any case, Meta fails to address that Brandstein contemplates a
`
`Griffiths-Jim beamformer in which a top path passes through speech and where “all
`
`other signals are attenuated,” and a bottom path that suppresses speech and where
`
`“all other signals are passed through.” In other words, the top path nulls out noise,
`
`and the bottom path nulls out speech. This sharply contrasts with the ’691 Patent
`
`invention where the first virtual microphone includes both speech and noise, such
`
`that a second virtual microphone with a substantially similar noise response may be
`
`3
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 9 of 31
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`used to cancel the noise and is exactly the type of “conventional array[] and
`
`algorithm[], which seek to reduce noise by nulling out noise sources” that the ’691
`
`Patent distinguishes.
`
`The Board should further deny the Petition because Petitioner fails to identify
`
`any reason or motivation that a POSITA would have to combine the multiple papers
`
`and embodiments within Brandstein on which the Petition relies to underpin its
`
`combination and for various limitations.
`
`II. THE ’691 PATENT
`U.S. Patent No. 8,503,691 (the “’691 Patent”) (Ex. 1001) discloses and claims
`
`methods for noise suppression in dual omnidirectional microphone arrays. Unlike
`
`“conventional arrays and algorithms, which seek to reduce noise by nulling out noise
`
`sources,” the ’691 Patent uses omnidirectional microphone arrays “to form two
`
`distinct virtual directional microphones which are configured to have very similar
`
`noise responses and very dissimilar speech responses.” ’691 Patent, Abstract. While
`
`“conventional multi-microphone systems attempt to increase the [signal-to-noise
`
`ratio] of the user’s speech by ‘steering’ the nulls of the system to the strongest noise
`
`sources[, t]his approach is limited in the number of noise sources removed by the
`
`number of available nulls.” Id., 1:46 -51. Counterintuitively, in the ’691 Patent,
`
`“[t]he only null formed by the DOMA is one used to remove the speech of the user
`
`from [the second virtual microphone]”. Id. at 4:67-5:2 (emphasis added). However,
`
`4
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 10 of 31
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`the similarity of the ’691 Patent’s virtual microphones’ noise responses allow them
`
`to be cancelled while leaving the speech response intact, “resulting in excellent noise
`
`suppression performance and minimal speech removal and distortion.” Id. at 8:1-3;
`
`10:25-42.
`
`The ’691 Patent discloses a noise suppression apparatus that creates two
`
`virtual microphones V1 and V2 based on the signals generated by two
`
`omnidirectional microphones.
`
`𝑉𝑉1(𝑧𝑧)=𝑂𝑂1(𝑧𝑧)𝑧𝑧−𝛾𝛾–𝛽𝛽𝑂𝑂2(𝑧𝑧)
`𝑉𝑉2(𝑧𝑧)= 𝑂𝑂2(𝑧𝑧)–𝑧𝑧−𝛾𝛾𝛽𝛽𝑂𝑂1(𝑧𝑧)
`𝛽𝛽=𝑑𝑑1𝑑𝑑2
`𝛾𝛾= 𝑑𝑑2− 𝑑𝑑1𝑐𝑐
` .𝑓𝑓𝑠𝑠 (samples)
`
`Id. at 10:55, 11:67. In these equations, β and γ are defined as follows:
`
`
`
`Id. at 10:58-62. “The distances d1 and d2 are the distances from O1 and O2 to the
`
`speech source,” respectively, and “γ is their difference divided by c, the speed of
`
`sound, and multiplied by the sampling frequency fs.” Id., 11:1-4.
`
`In the analog frequency domain, the equations for V1 and V2 are written as
`
`follows:
`
`
`
`
`
`5
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 11 of 31
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`
`. Ex. 2001, ¶ 30.
`
`
`
`
`
`where 𝑧𝑧=𝑒𝑒𝑗𝑗2𝜋𝜋𝜋𝜋and 𝜏𝜏= 𝑑𝑑2− 𝑑𝑑1𝑐𝑐
`
`The ’691 Patent discloses that the apparatus “works very well because the
`
`noise (far-field) responses of V1 and V2 are very similar while the speech (near-field)
`
`responses are very different.” ’691 Patent at 17:5-8.
`
`The ’691 Patent discloses polar plots depicting similar noise responses which
`
`may be cancelled while leaving dissimilar speech responses intact. For instance,
`
`Figures 10 and 12 show the linear responses of virtual microphones V2 and V1,
`
`respectively, to noise from a source at 1 meter:
`
`’691 Patent, 11:25-29, 12:25-29, FIGS. 10, 12.
`
`Figures 9 and 11 show the linear responses of virtual microphones V2 and V1,
`
`respectively, to speech at a distance of 0.1 meter at an angle of zero degrees:
`
`
`
`6
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 12 of 31
`
`
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`
`
`
`’691 Patent, 11:21-25, 12:20-24, FIGS. 9, 11.
`
`Figure 9 shows a null in the linear response of virtual microphone V2 to speech
`
`at zero degrees, “where the speech is typically expected to be located.” Id., 11:21-
`
`25. As shown in Figure 10, “[t]he linear response of V2 to noise is devoid of or
`
`includes no null, meaning all noise sources are detected.” Id., 11:25-29. This
`
`configuration ensures that virtual microphone V2 will detect all of the noise in front
`
`of the user so that it can be removed, which is an advantage over “conventional
`
`systems that can have difficulty removing noise in the direction of the mouth of the
`
`user.” Id., 11:38-40. In addition, “the superior noise suppression made possible using
`
`this system more than compensates for the initially poorer SNR,” or signal to noise
`
`ratio, of the virtual microphones. Id., 12:58-60.
`
`III. THE RELEVANT ALLEGED PRIOR ART
`A. Brandstein (Ex. 1005)
`Microphone Arrays: Signal Processing Techniques and Applications, (Ex.
`
`1005, “Brandstein”) is a collection of articles discussing topics in microphone
`
`7
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 13 of 31
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`arrays. Ex. 1005 at 5. Brandstein purports to have been published in 2001. Id. at 4.
`
`As discussed below, infra Section VI.B.2, Brandstein is not a single reference, it is
`
`a collection of different papers by different authors reproduced as a collection.
`
`Petitioner primarily relies on the paper “Robust Adaptive Beamforming” as
`
`reproduced in Brandstein as Chapter 5. See Brandstein at 77-89. Petitioner also relies
`
`on other papers for certain limitations and teachings for the proposed combination.
`
`See infra Section VI.B.2
`
`IV. CLAIM CONSTRUCTION
`For the purposes of this Preliminary Response only, Patent Owner believes
`
`that claim construction is not required to resolve any issues.
`
`V. LEVEL OF ORDINARY SKILL IN THE ART
`For the purposes of this Preliminary Response only, Patent Owner utilizes
`
`Petitioner’s proposed level of skill in the art—“a minimum of a bachelor’s degree in
`
`computer engineering, computer science, electrical engineering, mechanical
`
`engineering, or a similar field, and approximately three years of industry or academic
`
`experience in a field related to acoustics, speech recognition, speech detection, or
`
`signal processing. Work experience could substitute for formal education, and
`
`additional formal education could substitute for work experience” (citation omitted).
`
`Pet. at 18.
`
`8
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 14 of 31
`
`
`
`
`VI. ARGUMENT
`A. Legal Standard
`The question of obviousness is resolved on the basis of underlying factual
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`
`determinations, including: (1) the scope and content of the prior art, (2) any
`
`differences between the claimed subject matter and the prior art, (3) the level of skill
`
`in the art, and (4) where in evidence, so called secondary considerations. Graham v.
`
`John Deere Co. of Kansas City, 383 U.S. 1, 17–18 (1966). A claim is only
`
`unpatentable under 35 U.S.C. § 103 if “the differences between the subject matter
`
`sought to be patented and the prior art are such that the subject matter as a whole
`
`would have been obvious at the time the invention was made to a person having
`
`ordinary skill in the art to which said subject matter pertains.” KSR Int’l Co. v.
`
`Teleflex Inc., 550 U.S. 398, 406 (2007) (quoting 35 U.S.C. § 103).
`
`“An invention is not obvious simply because all of the claimed limitations
`
`were known in the prior art at the time of the invention. Instead, we ask ‘whether
`
`there is a reason, suggestion, or motivation in the prior art that would lead one of
`
`ordinary skill in the art to combine the references, and that would also suggest a
`
`reasonable likelihood of success.’” Forest Lab’ys, LLC v. Sigmapharm Lab’ys, LLC,
`
`918 F.3d 928, 934 (Fed. Cir. 2019). “Of course, concluding that the references are
`
`within the scope and content of the prior art to be considered for obviousness does
`
`not end the inquiry. Graham makes clear that the obviousness inquiry requires a
`
`9
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 15 of 31
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`determination whether the claimed invention would have been obvious to a skilled
`
`artisan.” Apple Inc. v. Samsung Elecs. Co., 839 F.3d 1034, 1050 n.14 (Fed. Cir.
`
`2016).
`
`B. Ground 1: Brandstein Does Not Render Obvious Claims 1-
`20
`1.
`
`Petitioner’s Combination Does Not Disclose or
`Render Obvious “the second linear response to noise
`[is/being] substantially similar to the first linear
`response to noise, and the second linear response to
`speech [is/being] substantially dissimilar to the first
`linear response to speech” as Recited in Independent
`Claims 1, 23, 27, 28, 29, and 41
`The Petition argues that Brandstein alone discloses this limitation based on
`
`the description related to the Griffiths-Jim beamformer (GJBF) in Chapter 5. Meta
`
`does not argue that this limitation is inherent or disclosed in view of any reference
`
`other than Brandstein.
`
`At its heart, Meta’s argument wrongly assumes that all systems that suppress
`
`noise with virtual microphones necessarily have “substantially similar” noise
`
`responses, ignoring contrary disclosure throughout Chapter 5 of Brandstein. As the
`
`Board aptly put it in rejecting this exact argument, “In effect, Petitioner is arguing
`
`that knowing that the sum of two numbers is equal to ’2’ necessarily means that each
`
`number is ’1’, as opposed, for example, to one number being ’2” and the other ’0’.
`
`IPR2023-01130, Paper 10 at 24. Moreover, as the Board further recognized, a simple
`
`10
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 16 of 31
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`examination of Fig. 5.3 “refutes Petitioner’s assertion [that] the virtual microphone
`
`noise responses are substantially similar.” Id.
`
`
`
`Ex. 2003, ¶ 45; Ex. 1005 at 90.
`
`The Board’s decision applies exactly to the arguments here. Among other
`
`findings, the Board found that Petitioner’s reliance on Brandstein’s Fig. 5.2, which
`
`shows the overall directivity of Branstein’s GJBF, in an attempt to infer the
`
`directivity of Brandstein’s individual virtual microphones, was misplaced:
`
`We agree with Patent Owner that the record does not
`support Petitioner’s assertion that Brandstein teaches or
`suggests the independent claim requirement, “wherein the
`first virtual microphone and the second virtual microphone
`are distinct virtual directional microphones with
`substantially similar responses to noise.” Contrary to
`Petitioner’s interpretation of Brandstein Figure 5.2, that
`figure does not lead to an inference that the individual
`microphone responses have similar responses to noise.
`Prelim. Resp. 12–13. In effect, Petitioner is arguing that
`knowing that the sum of two numbers is equal to “2”
`necessarily means that each number is “1”, as opposed, for
`example, to one number being “2” and the other “0”.
`
`11
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 17 of 31
`
`
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`
`
`Moreover, as discussed above, Brandstein’s description of
`Figure 5.1 demonstrates that the virtual microphones of
`Brandstein do not have “substantially similar responses to
`noise.” Prelim. Resp. 9–11. Likewise, Figure 5.3 of
`Brandstein also refutes Petitioner’s assertion that the
`virtual microphone noise responses are substantially
`similar. Prelim. Resp. 13.
`
`IPR2023-01130, Paper 10 at 24-25; see also Paper 12 at 5-6, 8 (reaffirming
`
`findings). Meta’s arguments rely on the same flawed interpretation of Brandstein’s
`
`Figure 5.2, and similarly fail to address Brandstein’s clear disclosure that its virtual
`
`microphones do not have substantially similar responses to noise that the Board
`
`noted in its earlier decision. Pet. at 32-34. Tellingly, Meta relies heavily on
`
`Brandstein’s Fig. 5.3 for the “speech response” limitations, but entirely fails to
`
`address its disclosure of very different noise responses in the context of the “noise
`
`response” limitations. Pet. at 34. Institution should be denied for the same reasons
`
`here.
`
`At the outset, as the Board found in IPR2023-01130, even if the outputs of the
`
`GJBF’s top and bottom branches are distinct virtual microphones, the Petition fails
`
`to identify any actual disclosure where their noise responses are substantially similar
`
`– there is none. Instead, the Petition’s GJBF-based arguments assume that any
`
`system with a virtual microphone that nulls speech, and another that does not,
`
`automatically satisfies the claim. But the speech response is only half of the
`
`12
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 18 of 31
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`limitation. Petitioner fails to show any disclosure wherein virtual microphones have
`
`“the second linear response to noise being substantially similar to the first linear
`
`response to noise.”
`
`The GJBF described in Brandstein has a very different design than the
`
`patented invention, because its top path is designed to isolate speech as much as
`
`possible, instead of capturing both speech and noise (such that the noise may be
`
`cancelled using another virtual microphone with a substantially similar noise
`
`response). Ex. 2001, ¶ 42. Brandstein clearly states that for the top path “[t]he [fixed
`
`beamformer] is designed to form a beam in the look direction so that the target signal
`
`is passed [through] and all other signals are attenuated,” while for the bottom path
`
`“the [blocking matrix] forms a null in the look direction so that the target signal is
`
`suppressed and all other signals are passed through.” Ex. 1005 at 78 (emphasis
`
`added). In other words, the fixed beamformer (top path) passes speech through (the
`
`target signal) and attenuates noise, while the blocking matrix (bottom path)
`
`suppresses speech and passes noise through. These are exactly the type of
`
`“conventional arrays and algorithms, which seek to reduce noise by nulling out noise
`
`sources” (i.e., by attenuating noise in the top path) distinguished by the ’691 Patent.
`
`’691 Patent at 4:63-67; Ex. 2001, ¶ 41. This approach is very different than the
`
`invention of the ’691 Patent, in which the first virtual microphone passes through
`
`both speech and noise, such that another virtual response with a “substantially
`
`13
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 19 of 31
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`similar” noise response may be used to cancel noise while preserving speech. A
`
`virtual microphone that attenuates noise plainly does not have a “substantially
`
`similar” noise response to one that passes noise through.
`
`Indeed, Petitioner admits that “the first virtual microphone captures the
`
`target speech signal, while the second virtual microphone blocks it.” Pet. at 7
`
`(emphasis added). Petitioner does not show any disclosure that the first virtual
`
`microphone is also designed to capture noise, and entirely fails to reconcile this
`
`position with the requirement for a substantially similar response to noise in both
`
`virtual microphones.
`
`Petitioner’s purported support for this limitation amounts to unsupported
`
`conclusions that any GJBF meets this limitation based on the dissimilarity of speech
`
`responses without substantively addressing the requirement for substantially similar
`
`noise responses. These arguments are substantively identical to those that the Board
`
`already rejected in IPR2023-01130.
`
`First, Petitioner argues that “[b]oth virtual microphones also have essentially
`
`the same . . . responses to noise, so that the noise responses cancel out in the GJBF’s
`
`final subtraction step.” Citing to Ex. 1003, ¶ 101. Pet. at 32. At the outset, Dr.
`
`Johnson’s opinions on this point are not entitled to any weight since they are
`
`conclusory and merely parrot the Petition. Compare id. with Pet. at 29. This
`
`argument amounts to an unsupported assumption that all systems in which noise is
`
`14
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 20 of 31
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`ultimately cancelled necessarily have virtual microphones with similar noise
`
`responses. Ex. 2001, ¶ 42. It also completely ignores Brandstein’s explanation that
`
`it uses a “multiple input canceller” (“MC”) which “generate[s] replicas of
`
`components correlated with the interferences” that are “subtracted from a delayed
`
`output signal [] of the fixed beamformer” based on which “the target signal is
`
`enhanced and undesirable signals such as ambient noise and interferences, are
`
`suppressed.” Ex. 1005 at 78-79. In other words, instead of using virtual microphones
`
`with substantially similar noise responses, Brandstein’s GJBF attempts to isolate
`
`speech in the top path as much as possible, then uses the MC to subtract only those
`
`noise components of the bottom path that are “correlated with the interferences.” Id.;
`
`see also Ex. 2001, ¶ 42. Petitioner’s speculation cannot stand in the face of
`
`Brandstein’s explicit teachings that it does not use substantially similar noise
`
`responses to cancel noise as Petitioner assumes, and its expert’s conclusory
`
`assertions do not cure that deficiency.
`
`Second, Petitioner argues that “as highlighted in red in Fig. 5.2, ’in the
`
`direction of the interference, a deep null is formed’ . . . ’[reflecting] that subtracting
`
`the bottom branch from the top branch cancels the interference, confirming that the
`
`virtual microphones’ responses to the interference are essentially the same,” citing
`
`Figure 5.2 which shows the directivity pattern for the final output of an example
`
`Griffiths-Jim beamformer:
`
`15
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 21 of 31
`
`
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`
`
`
`Pet. at 34. Figure 5.2 shows the overall directivity of an exemplary GJBF at three
`
`different frequencies (i.e., the combination of both the top and bottom path), not the
`
`directivity of either virtual microphone individually. That directivity merely has an
`
`area of significantly reduced gain in the direction of an interference (i.e., a noise
`
`source). Ex. 2001, ¶ 43. This reflects the overall result of the system in attenuating
`
`noise; it discloses nothing about the comparative responses to noise of the top or
`
`bottom path within that system (which are not even depicted in the Figure). Id. In
`
`any case, contrary to Petitioner’s speculation, Brandstein contemplates this result
`
`based on subtracting elements of noise (generated by a multiple input canceller by
`
`taking “replicas of components correlated with the interferences”) from an FBF
`
`signal that already has as little noise as possible, not based on any “substantially
`
`similar” noise response. Id.
`
`Moreover, while speculating about the noise responses based on the overall
`
`output, Petitioner does not address contradictory disclosure in Figure 5.3
`
`immediately following the cited portion. This is particularly telling since Petitioner
`
`16
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 22 of 31
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`relies heavily on Figure. 5.3 for showing dissimilarity of speech responses in the
`
`same section. Pet. at 33-34. Unlike Figure 5.2, Figure 5.3 actually does depict the
`
`directivity of the fixed beamformer (top path) and blocking matrix (bottom path) in
`
`relation to a target signal (i.e., speech). Ex. 2001, ¶ 44. A POSITA would plainly
`
`understand that their noise responses (highlighted in green) are not substantially
`
`similar.
`
`
`
`Ex. 2001, ¶ 44; Ex. 1005 at 80. As Brandstein’s Chapter 5 does not disclose any
`
`equations with which to determine the responses of its top and bottom branches, Fig.
`
`5.3 is the only disclosure of those responses outside the above-cited discussion that
`
`noise in the FBF (top branch) is attenuated compared to the BM (bottom branch).
`
`Ex. 2001, ¶ 46.
`
`While Petitioner relies on an annotated version of Fig. 5.3 in the context of
`
`the “speech response” terms, it tellingly ignores the same figure when discussing
`
`the “noise response” limitations. Compare Pet. at 33 with Pet. at 34.
`
`17
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 23 of 31
`
`
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`
`
`
`The sum total of Meta’s discussion of Fig. 5.3 as to noise—not noise response—is a
`
`single red line in Fig. 5.3 at approximately 40 degrees. This is, at best, mere hand
`
`waving. Meta does not even assert that responses to noise in the 40-degree direction
`
`are substantially similar; presumably because Brandstein’s disclosed gain chart
`
`would directly contradict it. See IPR2023-01130, Paper 10 at 24 (“Figure 5.3 of
`
`Brandstein also refutes Petitioner’s assertion [that] the virtual microphone noise
`
`responses are substantially similar” (citing Prelim. Resp. at 13)). In any case, the
`
`noise responses of the FBF and BM in Fig. 5.3 are plainly very different, as shown
`
`in Dr. Sayeed’s annotations above. Ex. 2001, ¶ 45.
`
`
`
`Because the Petition fails to show any disclosure of “the second linear
`
`response to noise [is/being] substantially similar to the first linear response to noise,
`
`and the second linear response to speech [is/being] substantially dissimilar to the
`
`18
`
`
`
`Case 6:23-cv-00158-ADA Document 76-5 Filed 05/08/24 Page 24 of 31
`
`
`
` IPR2024-00349
` PATENT NO. 8,503,691
`
`
`first linear response to speech” as required by every independent claim, and should
`
`therefore be denied.
`
`2.
`
`Petitioner Does Not Identify Any Motivation to
`Combine Embodiments and References Within
`Brandstein
`Brandstein (Ex. 1005) is not a single reference. It is a collection of different
`
`papers by different authors, each reproduced as a chapter in Brandstein. Each paper
`
`is a separate reference, and contains separate embodiments, for which the Petitioner
`
`bears the burden of showing a reason or motivation to combine.