`Tel: 571-272-7822
`
`
`Paper 28
`Entered: July 12, 2018
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`_______________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`_______________
`
`APPLE INC.,
`Petitioner,
`
`v.
`
`ANDREA ELECTRONICS CORP.,
`Patent Owner.
`_______________
`
`Case IPR2017-00626
`Patent 6,363,345 B1
`_______________
`
`
`
`Before STEPHEN C. SIU, MICHAEL R. ZECHER, and
`JEREMY M. PLENZLER, Administrative Patent Judges.
`
`PLENZLER, Administrative Patent Judge.
`
`
`
`
`FINAL WRITTEN DECISION
`Inter Partes Review
`35 U.S.C. § 318 and 37 C.F.R. § 42.73
`
`
`
`
`
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`
`I. INTRODUCTION
`We have jurisdiction to hear this inter partes review under 35 U.S.C.
`§ 6, and this Final Written Decision is issued pursuant to 35 U.S.C. § 318(a)
`and 37 C.F.R. § 42.73. For the reasons that follow, we determine that Apple
`Inc. (“Petitioner”) has shown by a preponderance of the evidence that claims
`1–3, 12–25, 38, and 47 of U.S. Patent No. 6,363,345 B1 (Ex. 1001, “the
`’345 patent”) are unpatentable, but has failed to establish that claims 4–11
`and 39–46 of the ’345 patent are unpatentable.
`A. Background
`Petitioner filed a Petition to institute an inter partes review of claims
`1–25 and 38–47 the ’345 patent. Paper 1 (“Pet.”). Andrea Electronics Corp.
`(“Patent Owner”) filed a Preliminary Response. Paper 6 (“Prelim. Resp.”).
`Pursuant to 35 U.S.C. § 314, we instituted trial on all challenged
`claims, and on all challenges raised in the Petition. See Paper 7 (“Dec. to
`Inst.”), 12–13. After institution of trial, Patent Owner filed a Patent Owner
`Response (Paper 11, “PO Resp.”), to which Petitioner filed a Reply (Paper
`18, “Pet. Reply”).
`An oral argument was held on April 25, 2018. A transcript of the oral
`argument is included in the record. Paper 25 (“Tr.”).
`B. Related Matters
`Petitioner and Patent Owner identify a number of proceedings, both in
`district court and before the Patent Trial and Appeal Board, involving
`patents related to the ’345 patent, including a district court proceeding
`specifically directed to the ’345 patent with Petitioner as a party and Case
`IPR2017-00627, which is directed to the ’345 patent and involves the same
`
`2
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`parties as this proceeding. Pet. viii–x; Paper 4, 1. Our final decision in Case
`IPR2017-00627 issues concurrently herewith.
`C. Asserted Grounds of Unpatentability and Evidence of Record
`Petitioner contends that the challenged claims are unpatentable under
`35 U.S.C. §§ 102 and 103 as set forth in the table below (Pet. 2–3, 17–69).
`Reference(s)
`Basis Claims Challenged
`Hirsch1
`§ 102
`1–3, 12, 13, 21, 23, and 38
`
`Hirsch and Martin2
`
`Hirsch and Boll3
`
`§ 103
`
`4–11, 25, 39–42, and 46
`
`§ 103
`
`13, 14, 17–21, 23, and 474
`
`Hirsch, Martin, and Boll
`
`§ 103
`
`43
`
`Hirsch, Boll, and Arslan5
`
`§ 103
`
`15, 16, and 24
`
`Hirsch and Uesugi6
`
`§ 103
`
`22
`
`Hirsch, Martin, and Uesugi
`
`§ 103
`
`44 and 45
`
`Petitioner provides testimony from Bertrand Hochwald, Ph.D.
`Ex. 1003; Ex. 1023. Patent Owner provides testimony from Scott C.
`Douglas, Ph.D. Ex. 2002. Petitioner also provides deposition testimony
`
`1 H.G. Hirsch & C. Ehrlicher, “Noise Estimation Techniques for Robust
`Speech Recognition,” IEEE 1995 (Ex. 1005, “Hirsch”).
`2 Ranier Martin, “An Efficient Algorithm to Estimate the Instantaneous SNR
`of Speech Signals,” Eurospeech 1993 (Ex. 1006, “Martin”).
`3 Steven F. Boll, “Suppression of Acoustic Noise in Speech Using
`Spectral Subtraction,” IEEE 1979 (Ex. 1009, “Boll”).
`4 Dependent claim 22 is included in this asserted ground of unpatentability
`in Petitioner’s “Identification of Claims being Challenged (§ 42.104(b))”
`(Pet. 3) (emphasis omitted), but this claim is not addressed in its substantive
`analysis (id. at 49–52).
`5 U.S. Patent No. 5,706,395, iss. Jan. 6, 1998 (Ex. 1011, “Arslan”).
`6 U.S. Patent No. 5,459,683, iss. Oct. 17, 1995 (Ex. 1015, “Uesugi”).
`
`3
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`from Dr. Douglas (Ex. 1026) and Patent Owner provides deposition
`testimony from Dr. Hochwald (Ex. 2005).
`D. The ’345 Patent
`The ’345 patent “relates to noise cancellation and reduction and, more
`specifically, to noise cancellation and reduction using spectral subtraction.”
`Ex. 1001, 1:19–21. The ’345 patent explains that its system receives a noise
`signal and converts that signal to the frequency domain through a Fast
`Fourier Transform (FFT). Id. at 4:50–5:14. Separate thresholds are set for
`each frequency bin to determine the location of noise elements for each
`frequency bin separately. Id. at 6:10–13. The ’345 patent determines the
`thresholds by setting two minimum values, which are described as a future
`minimum and a current minimum. Id. at 6:23–41.
`At predetermined time intervals (e.g., every 5 seconds), the future
`minimum value is initialized as the value of the current magnitude of the
`signal. Id. at 6:24–28. Over that time interval, and before the next
`initialization, the future minimum value of each bin is compared with the
`current magnitude value of the signal. Id. If the current magnitude is
`smaller than the future minimum, the value of the future minimum is
`replaced with that current magnitude. Id. at 6:28–32.
`At the start of each time interval, the current minimum is set as the
`value of the future minimum that was determined over the previous time
`interval. Id. at 6:34–38. The current minimum then follows the minimum
`value of the signal for the next time interval by comparing its value with the
`current magnitude value. Id. The current minimum value is used by the
`spectral subtraction process to remove noise from the signal. Id. at 6:38–41.
`
`4
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`
`E. Illustrative Claims
`As noted above, Petitioner challenges claims 1–25 and 38–47 of the
`’345 patent. Claims 1 and 38 are independent, with claims 2–25 and 39–47
`depending from either claim 1 or 38. Claim 1 is illustrative, and is
`reproduced below:
`1. An apparatus for canceling noise, comprising:
`an input for inputting an audio signal which includes a noise
`signal;
`a frequency spectrum generator for generating the frequency
`spectrum of said audio signal thereby generating frequency
`bins of said audio signal; and
`a threshold detector for setting a threshold for each frequency bin
`using a noise estimation process and for detecting for each
`frequency bin whether the magnitude of the frequency bin is
`less than the corresponding threshold, thereby detecting the
`position of noise elements for each frequency bin.
`Ex. 1001, 9:35–46.
`
`
`II. ANALYSIS
`A. Claim Construction
`“[W]e need only construe terms ‘that are in controversy, and only to
`the extent necessary to resolve the controversy.’” Nidec Motor Corp. v.
`Zhongshan Broad Ocean Motor Co., 868 F.3d 1013, 1017 (Fed. Cir. 2017)
`(quoting Vivid Techs., Inc. v. Am. Sci. & Eng’g, Inc., 200 F.3d 795, 803
`(Fed. Cir. 1999)). We construe the claims using the broadest reasonable
`construction in light of the ’345 patent Specification. See 37 C.F.R.
`§ 42.100(b). Applying that standard, we interpret the claim terms of the
`’345 patent according to their ordinary and customary meaning in the
`context of the patent’s written description. See In re Translogic Tech., Inc.,
`
`5
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`504 F.3d 1249, 1257 (Fed. Cir. 2007). An inventor is entitled to be his or
`her own lexicographer of patent claim terms by providing a definition of the
`term in the specification with reasonable clarity, deliberateness, and
`precision. In re Paulsen, 30 F.3d 1475, 1480 (Fed. Cir. 1994). In the
`absence of such a definition, however, limitations are not to be read from the
`specification into the claims. In re Van Geuns, 988 F.2d 1181, 1184 (Fed.
`Cir. 1993).
`For purposes of this decision, we determine that no term requires
`express construction to resolve the controversies between the parties
`identified below.
`
`B. Challenges
` Hirsch
`Petitioner challenges claims 1–3, 12, 13, 21, 23, and 38 as anticipated
`by Hirsch under 35 U.S.C. § 102. Pet. 22–31. “In [its] Response, Patent
`Owner does not address the anticipation grounds with respect to the Hirsch
`reference.” PO Resp. 12. Based on our review of the record before us, we
`are persuaded that Petitioner has established by a preponderance of the
`evidence that claims 1–3, 12, 13, 21, 23, and 38 are anticipated by Hirsch.
`Claim 1, for example, is directed to “[a]n apparatus for canceling
`noise.” Claim 1 requires an apparatus that receives an audio signal (“an
`input for inputting an audio signal”), converts that audio signal to the
`frequency domain to provide frequency bins (“a frequency spectrum
`generator for generating the frequency spectrum of said audio signal”), and
`determines that the various frequency bins of that signal are noise if their
`magnitudes are below corresponding thresholds (“a threshold detector for
`setting a threshold for each frequency bin . . . and for detecting . . . whether
`
`6
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`the magnitude of the frequency bin is less than the corresponding threshold,
`thereby detecting the position of noise elements for each frequency bin”).
`Claim 38 is similar to claim 1, but is directed to a “method” rather than an
`“apparatus,” and additionally recites “subtracting said noise elements . . .
`from said audio signal.” We adopt Petitioner’s unrebutted arguments and
`evidence related to how Hirsch properly accounts for each and every
`element of claims 1 and 38. See Pet. 18–31.
`Hirsch is directed to “techniques . . . to estimate the noise spectra or
`the noise characteristics for noisy speech signals” that “can be combined
`with a nonlinear spectral subtraction scheme.” Ex. 1005, 153. Hirsch
`discusses use of “FFT based spectral analysis” (id. at 154), for example,
`which Petitioner cites as corresponding to the “generating the frequency
`spectrum of said audio signal thereby generating frequency bins of said
`audio signal” recited in claim 1 (Pet. 27). Hirsch discusses use of an
`“adaptive threshold” for each frequency bin to distinguish between speech
`and noise (Ex. 1005, 153), which Petitioner cites as corresponding to the
`“setting a threshold for each frequency bin using a noise estimation process”
`recited in claim 1 (Pet. 27). Hirsch describes comparing spectral
`components to their corresponding thresholds to determine if speech is
`present (Ex. 1005, 153), which Petitioner cites as corresponding to the
`“detecting for each frequency bin whether the magnitude of the frequency
`bin is less than the corresponding threshold, thereby detecting the position of
`noise elements for each frequency bin” recited in claim 1 (Pet. 27–28).
`Based on the record before us, we are persuaded that Hirsch discloses each
`and every element of claim 1.
`
`7
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`
`We also are persuaded that Hirsch discloses each and every element
`of claim 38 for the reasons discussed above in connection with claim 1, as
`well as the “subtracting said noise elements . . . from said audio signal”
`additionally recited in claim 38. With respect to the recited “subtracting,”
`Petitioner cites Hirsch’s discussion of spectral subtraction. Pet. 25–26
`(citing Ex. 1005, 153–155). As Petitioner notes, “[t]he spectral subtraction
`technique shown in Hirsch subtracts the noise estimate from each sub-band
`or ‘bin’ and sets any negative values to zero” (id. (citing Ex. 1005, 153)),
`which “is the same technique used in the ’345 patent” (id. (citing Ex. 1001,
`6:58–61, 7:16–27)).
`Claims 2, 3, 12, 13, 21, and 23 depend from claim 1, and Petitioner
`cites to relevant portions of Hirsch as disclosing the additional limitations of
`those claims. See Pet. 28–31. We are persuaded by Petitioner’s unrebutted
`arguments and evidence related to how Hirsch properly accounts for each
`and every element of these dependent claims, which we adopt for purposes
`of this decision. Specifically, we find that Hirsch discloses the additional
`limitations recited in those claims based on the citations to Hirsch therein
`and the supporting expert testimony. See id.
`Based on our review of the record before us, the preponderance of the
`evidence supports Petitioner’s contentions that claims 1–3, 12, 13, 21, 23,
`and 38 are anticipated by Hirsch.
` Hirsch and Martin
`Petitioner challenges claims 4–11, 25, 39–42, and 46 as unpatentable
`over Hirsch and Martin under 35 U.S.C. § 103. Pet. 38–49. Patent Owner
`responds that “Petitioner fails to demonstrate by a preponderance of the
`evidence that claims 4–11 and 39–42 are unpatentable as alleged,” but does
`
`8
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`not address specifically Petitioner’s contentions regarding claim 25 or claim
`46. PO Resp. 13. Based on our review of the record before us, Petitioner
`has established that claim 25 is unpatentable over Hirsch and Martin, but has
`failed to establish by a preponderance of the evidence that claims 4–11, 39–
`42, and 46 are unpatentable over Hirsch and Martin.
`a. Claims 4–11, 39–42, and 46
`Claim 4 depends from claim 1, and claims 5–11 ultimately depend
`from claim 4. Claim 39 depends from claim 38, and claims 40–42 and 46
`ultimately depend from claim 39. Claims 4 and 39 each recite that “the
`threshold for each frequency bin [is set] in accordance with a current
`minimum value of the magnitude of the corresponding frequency bin” and
`“said current minimum value [is] derived in accordance with a future
`minimum value of the magnitude of the corresponding frequency bin.” With
`respect to claims 4 and 39, Petitioner acknowledges that “Hirsch does not
`expressly disclose . . . maintaining the noise threshold for each frequency bin
`using ‘current minimum’ and ‘future minimum’ values,” but contends that
`these features are “obvious variations of Hirsch in view of the guidance in
`Martin.” Pet. 31–32 (emphases omitted).
`Martin is directed to “estimat[ing] the instantaneous signal-to-noise
`ratio of speech signals” (Ex. 1006, 1093), and is specifically referenced in
`Hirsch as a known way to estimate noise characteristics in a past speech
`segment (Ex. 1005, 153). Petitioner provides an annotated version of
`Martin’s Figure 2 (Pet. 39), reproduced below.
`
`9
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`
`
`This annotated version of Martin’s Figure 2 is a flow chart illustrating a
`signal-to-noise ratio estimation algorithm (Ex. 1006, 1094), and Petitioner’s
`annotations include colored boxes around certain portions of the flow chart
`for purposes of discussion. Petitioner cites Martin’s noise floor estimate
`Pn(i) as corresponding to the “current minimum value” and Martin’s
`minimum observed noise level during a period PMmin as corresponding to the
`“future minimum value” recited in claims 4 and 39. Pet. 41–42. Petitioner
`explains that, “[d]uring each period of M samples, Martin’s algorithm tracks
`the observed power level of the signal and stores the minimum value as
`PMmin (red box),” and “[a]t the end of the period (purple box), the algorithm
`sets the estimated noise floor Pn(i) to the minimum observed power value
`(right-hand side blue box).” Pet. 39 (citing Ex. 1006, 1094). The Petition
`relies on a specific scenario where “W=1 (the number of sub-windows)” in
`its contentions regarding the teachings of Martin. Pet. 40.
`
`10
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`
`Dr. Hochwald provides an example illustrating the algorithm of
`Martin based on the scenario with W = 1, which is reproduced below.
`Ex. 1003 ¶¶ 137–140.
`
`
`
`The figure reproduced above is a graphical illustration of the determination
`of current magnitude, future minimum, and current minimum with Martin’s
`Figure 2 algorithm. Dr. Hochwald explains that, “[i]n response to [the
`current magnitude] data stream, Martin’s algorithm would derive the values
`for the estimated noise power, Pn(i), and the minimum observed noise
`power, PMmin, that are listed in the figure above,” with “[t]he arrows in the
`chart above show[ing] how values change when the window of M samples
`ends.” Ex. 1003 ¶ 137.
`Petitioner contends that, “[w]hile Martin shows its algorithm
`operating on signal power, it teaches that the algorithm can be adapted to
`work with spectral subtraction techniques that operate on the magnitude of a
`signal.” Pet. 40 (citing Ex. 1006, 1096). Petitioner further contends that “a
`person of ordinary skill in the art would have understood that Martin’s
`
`11
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`algorithm also could be used to track the magnitude of the noise signal
`instead of its power.” Id. at 34 (emphases omitted) (citing Ex. 1003 ¶ 123).
`Dr. Hochwald testifies that “[t]he power of an audio signal is its magnitude
`squared” and “a person of ordinary skill can readily move between the two.”
`Ex. 1003 ¶ 123.
`Petitioner contends, for example, that a person skilled in the art would
`have combined the teachings of Martin with those of Hirsch “because Hirsch
`not only cites to Martin, but identifies relevant benefits it provides in
`spectral subtraction schemes.” Pet. 34 (citing Ex. 1005, 156; Ex. 1006,
`1096; Ex. 1003 ¶¶ 128, 129). Petitioner reasons that it would have been
`obvious to modify Hirsch’s teachings “to incorporate Martin’s technique for
`estimating the noise floor of a signal within Hirsch’s adaptive threshold
`calculation to improve operation of Hirsch’s system in non-stationary noise
`settings.” Id. at 35 (citing Ex. 1003 ¶¶ 129–32).
`Patent Owner responds that “[Petitioner] and its expert, Dr.
`Hochwald, analyze the combination of Martin and Hirsch where W =1, i.e.,
`where subwindows are eliminated in an apparent attempt to overcome the
`deficiencies of Martin.” PO Resp. 44 (underlining omitted). Patent Owner
`explains that “[a] key aspect of Martin’s algorithm is its ability to distinguish
`between slowly varying and rapidly varying noise levels – a distinction
`made by the monotonic decision block” and, “[a]s explained by Dr. Douglas,
`one of ordinary skill in the art would not have eliminated subwindows from
`the Martin system, because it would be directly contrary to Martin’s
`disclosure and it would thwart the ability of the algorithm to track rapidly
`varying noise power.” Id. (underlining omitted) (citing Ex. 1006, 194;
`
`12
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`Ex. 2002 ¶¶ 94–95). Based on the record developed during trial, Patent
`Owner’s contentions are persuasive.
`Initially, we agree with Petitioner that one skilled in the art would
`have considered Martin’s teachings, generally, when reviewing the teachings
`of Hirsch, as Martin is specifically referenced in Hirsch itself. Nevertheless,
`we are not persuaded that one skilled in the art would have modified Hisch’s
`system based on the teachings of Martin in the particular manner proposed
`by Petitioner.
`Consistent with Patent Owner’s contentions noted above, Martin
`describes determining “[t]he noise power estimate is based on the minimum
`of signal power within a window of L samples” and “[f]or reasons of
`computational complexity and delay the data window of length L is
`decomposed into W windows of length M such that M * W = L,” and
`provides an example where W=4. Ex. 1006, 1094. Martin distinguishes
`between “slowly varying noise power” and “rapidly varying noise power” to
`determine its noise power estimate, with a determination of whether the
`minimum power in successive sub-windows is monotonically increasing
`central to its estimation. Id. “If the minimum power of the last W windows
`with M samples each is monotonically increasing we decide on rapid noise
`power variation” and “the noise power estimate equals the power minimum
`of the last M samples Pn(i) = PMmin(i = r * M).” Id. “In case of non
`monotonic power the noise power estimate is set to the minimum of the
`length L window.” Id.
`Regardless of whether the noise power calculation is that for “slowly
`varying noise power” or “rapidly varying noise power,” Martin requires that
`first a determination is made as to whether or not “the minimum power of
`
`13
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`the last W windows with M samples each is monotonically increasing.” Id.
`(emphasis added). To the extent there can be any dispute on this issue, the
`testimony from Dr. Douglas supports the understanding that at least two
`subwindows are required to determine which noise power calculation to use.
`See Ex. 2002 ¶¶ 95, 96.
`In its Reply, Petitioner contends that, “[i]n this configuration [where
`W = 1], any purported distinction between windows and sub-windows in
`Martin vanishes and [Patent Owner]’s argument is irrelevant,”
`acknowledging that the proposed modification eliminates Martin’s sub-
`windows. Pet. Reply 5. Petitioner further contends that Patent Owner
`“incorrectly argues that the use of sub-windows is central to Martin’s
`algorithm” and “[n]othing in Martin suggests that the use of sub-windows is
`required.” Id. at 7 (citing Ex. 1023 ¶¶ 6–8). We disagree with Petitioner,
`and credit the testimony of Dr. Douglas over that of Dr. Hochwald with
`respect to Martin’s use of sub-windows because, as discussed below, Dr.
`Hochwald’s testimony is not consistent with the disclosure of Martin, while
`the testimony of Dr. Douglas is consistent with that disclosure. See, e.g.,
`Ex. 1023 ¶¶ 6–9 (asserting that Martin does not expressly require sub-
`windows); Ex. 2002 ¶¶ 94–96 (discussing Martin’s use of at least two sub-
`windows for monotonic determination).
`As noted above, Martin’s algorithm requires a determination of
`whether or not “the minimum power of the last W windows with M samples
`each is monotonically increasing.” Ex. 1006, 1094. We are not persuaded
`by Petitioner’s argument that one skilled in the art would have found it
`obvious to modify Hirsch’s system based on a scenario from Martin where
`W = 1 because such a scenario is counter to the entire purpose of Martin.
`
`14
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`Even Petitioner’s Declarant, Dr. Hochwald, acknowledges that a single data
`point (i.e., W = 1) cannot be used to determine whether something is
`monotonically increasing. See Ex. 2005 69:10–17 (explaining that you need
`at least two data points to make such a determination, meaning W ≥ 2).
`Petitioner fails to provide persuasive reasoning as to why one skilled in the
`art would have modified Hirsch’s teachings in a manner that is contrary to
`the express disclosure of Martin.
`Petitioner attempts to cure these deficiencies in its Reply, contending
`that, “[e]ven where Martin’s algorithm is configured to use multiple sub-
`windows, it still teaches the claimed ‘future minimum.’” Pet. Reply 8. We
`do not consider this new argument, as it is a new theory of unpatentability
`presented for the first time in the Petitioner’s Reply. See Dell Inc. v.
`Acceleron, LLC, 818 F.3d 1293, 1301 (Fed Cir. 2016) (“A patent owner . . .
`is undoubtedly entitled to notice of and a fair opportunity to meet the
`grounds of [unpatentability].”) (quoting Belden Inc. v. Berk-Tek LLC, 805
`F.3d 1064, 1080 (Fed. Cir. 2015)).
`For at least these reasons, Petitioner has failed to establish by a
`preponderance of the evidence that claims 4–11, 39–42, and 46 are
`unpatentable over the combined teachings of Hirsch and Martin.
`b. Claim 25
`Claim 25 depends from claim 1, and additionally recites “an adaptive
`array comprising a plurality of microphones for receiving said audio signal.”
`Petitioner cites Martin’s microphone arrays that include adaptive processing
`of microphone signals as teaching this feature. Pet. 48 (citing Ex. 1006,
`1096). Petitioner reasons that “[i]t would have been obvious to add these
`conventional adaptive microphone arrays to Hirsch’s speech processing
`
`15
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`systems.” Id. (citing Ex. 1003 ¶¶ 125–32). Petitioner explains that “[a]
`skilled person would have recognized such arrays to have been one of a
`small number of conventional design choices for capturing a signal, and
`could have integrated this conventional component into Hirsch’s system
`with only routine effort.” Id. at 38 (citing Ex. 1003 ¶ 145). Petitioner
`reasons that “a skilled person would have considered Hirsch together with
`Martin because Hirsch not only cites to Martin, but identifies relevant
`benefits it provides,” and “[a] skilled person also would have recognized that
`techniques such as those shown in both Hirsch and Martin are routinely
`combined.” Id. at 34–35. Patent Owner provides no evidence or argument
`to rebut Petitioner’s contentions and evidence in connection with claim 25.7
`Based on the record before, Petitioner’s unrebutted arguments and
`evidence establish by a preponderance of the evidence that it would have
`been obvious to modify Hirsch’s system to include “a plurality of
`microphones for receiving said audio signal,” as taught by Martin.
`Accordingly, Petitioner has established by a preponderance of the evidence
`that claim 25 is unpatentable over the combined teachings of Hirsch and
`Martin.
`
` Hirsch and Boll
`Petitioner challenges claims 13, 14, 17–21, 23, and 47 as unpatentable
`over the combined teachings of Hirsch and Boll under 35 U.S.C. § 103.
`Pet. 49–59. Claim 13 depends from claim 1, and further recites “a subtractor
`
`
`7 Patent Owner provides arguments that address, generally, Petitioner’s
`proposed combination of Martin’s teachings with those of Hirsch, but does
`not address specifically any issues in connection with claim 25. See
`generally PO Resp. 33–43. Patent Owner’s general arguments are directed
`to Martin’s algorithm rather than its use of multiple microphones.
`
`16
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`for subtracting said noise elements estimated at said positions determined by
`said threshold detector from said audio signal to derive said audio signal
`substantially without said noise.” Claims 14 and 17–20 each ultimately
`depend from claim 13. Claim 21 depends from claim 1, and further recites
`“an estimator for estimating a magnitude of each frequency bin.” Claim 23
`depends from claim 21. Claim 47 depends from claim 38 and recites
`limitations similar to those in claim 13. Petitioner cites Boll as teaching the
`additional features recited in those claims, and contends that one skilled in
`the art would have found it obvious to apply those teachings in Hirsch’s
`system. Pet. 49–59. Patent Owner does not dispute that Boll teaches the
`additional features recited in those claims, but responds that Petitioner has
`failed to provide a reason why one skilled in the art would have included
`those particular features in Hirsch’s system. PO Resp. 52–54.
`We are persuaded by Petitioner’s contentions regarding how Boll
`teaches the features in claims 13, 14, 17–21, 23, and 47, and find that the
`portions of Boll cited in the Petition as corresponding to those features teach
`the additional limitations found in claims 13, 14, 17–21, 23, and 47. We
`also are persuaded by Petitioner’s reasons why one of ordinary skill in the
`art would have combined Boll’s teachings with those of Hirsch. For
`example, Petitioner reasons that “a skilled person would have considered
`Hirsch in conjunction with Boll” because “Hirsch is focused on a particular
`improvement to the noise estimation process used in spectral subtraction,
`and logically does not provide details of the other aspects of the spectral
`subtraction process,” and “[a] skilled person . . . would have known that
`certain operations are ordinarily performed to carry out spectral subtraction,
`and would have looked to a reference like Boll for details in doing so.” Pet.
`
`17
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`51 (underlining omitted) (citing Ex. 1003 ¶ 147; Ex. 1005, 153, Abstract;
`Ex. 1009, 114). More specifically, Petitioner reasons that one skilled in the
`art would have found it obvious to use Boll’s well-known spectral
`subtraction features, such as filter multiplication, in Hirsch’s spectral
`subtraction process, as well as Boll’s other teachings including smoothing
`signal value inputs and residual noise reduction techniques. Id. at 51–52
`(citing Ex. 1003 ¶¶ 149, 150, 154, 156, 157).
`Patent Owner responds, generally, that Petitioner has failed to provide
`sufficient explanation as to why one skilled in the art would have made the
`particular modifications to Hirsch that it proposes. PO Resp. 52–54. Patent
`Owner does not rebut Petitioner’s contentions with respect to the teachings
`of Hirsch or Boll relied on in Petitioner’s reasoning for combining the
`teachings of those references or the supporting testimony from Dr.
`Hochwald. We credit the testimony of Petitioner’s declarant, Dr. Hochwald,
`in connection with these challenges, as that testimony is unrebutted by any
`testimony from Patent Owner. See Ex. 1003 ¶¶ 147, 149, 150, 154, 156,
`157. Indeed, Patent Owner provides no testimony from its declarant or
`deposition testimony from Petitioner’s declarant in response to these
`challenges. We are persuaded by Petitioner’s rationale and supporting
`evidence, which we adopt for purposes of this decision.
`There is no dispute that Hirsch teaches using spectral subtraction and,
`itself, states that it “can be combined with well-known spectral subtraction
`techniques” or that Boll is a well-known paper in the area of spectral
`subtraction. Ex. 1005, 156; see also Pet. 30 (contending that Hirsch
`discloses the features of claim 13, which Patent Owner does not dispute).
`Indeed, both Dr. Hochwald and Dr. Douglas provide testimony to that effect.
`
`18
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`Ex. 1003 ¶¶ 147, 149, 150, 154, 156,157; Ex. 1030, 17:1–8; 195:11–196:7,
`197:20–198:10. For example, Patent Owner’s declarant, Dr. Douglas,
`explains that Boll “is a foundational reference . . . on the general set of
`techniques associated with spectral subtraction.” Ex. 1030, 195:21–196:1.
`We are persuaded that selecting various ones of a finite number of known
`techniques for implementing spectral subtraction in Hirsch’s system, such as
`Boll’s filter multiplication, smoothing techniques, or residual noise
`reduction techniques, would have been obvious based on the unrebutted
`evidence before us.
`Accordingly, Petitioner has established by a preponderance of the
`evidence that claims 13, 14, 17–21, 23, and 47 are unpatentable over the
`combined teachings of Hirsch and Boll.
` Hirsch, Martin, and Boll
`Claim 43 depends from claim 40, which depends from claim 39.
`Petitioner’s contentions regarding claim 43 do not cure the deficiencies in
`either Hirsch or Martin with respect to the challenge to claim 39 identified
`above. See Pet. 59–60. Accordingly, Petitioner has failed to establish by a
`preponderance of the evidence that claim 43 is unpatentable over the
`combined teachings of Hirsch, Martin, and Boll.
` Hirsch, Boll, and Arslan
`Claim 15 depends from claim 13, claim 16 depends from claim 15,
`and claim 24 depends from claim 23. Those claims define additional
`features of the particular spectral subtraction techniques recited in the claims
`from which they depend, such as specifying the type of filter function for the
`spectral subtraction (claims 15 and 16) and the smoothing process (claim
`24). Petitioner cites Arslan as teaching the additional features in claims 15
`
`19
`
`
`
`IPR2017-00626
`Patent 6,363,345 B1
`
`and 16, as well as smoothing over neighboring frequencies in connection
`with claim 24, which we find persuasive, and adopt for purposes of this
`decision. Pet. 62–66. Patent Owner does not dispute those contentions.
`Rather, similar to the response to Petitioner’s challenge based on the
`combination of Hirsch and Boll, discussed above, Patent Owner responds,
`generally, that Petitioner “fails to establish that a skilled artisan would have
`been motivated to combine Hirsch and Boll,” but even if it had, Petitioner
`“makes no showing as to why a skilled artisan would have been motivated to
`further modify these references in view of Arslan.” PO Resp. 56
`(underlining omitted). We disagree with Patent Owner’s characterization of
`Petitioner’s contentions, which we find persuasive as explained below.
`Further, we credit the testimony of Petitioner’s declarant in support of these
`challenges, as that testimony is unrebutted by any testi