`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`____________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`____________________
`
`
`
`
`
`
`
`
`
`APPLE INC.
`Petitioner,
`
`v.
`
`ANDREA ELECTRONICS INC.,
`Patent Owner.
`
`Patent No. 6,363,345
`Issued: March 26, 2002
`Filed: February 18, 1999
`Inventors: Joseph Marash, et al.
`Title: SYSTEM, METHOD AND APPARATUS FOR CANCELLING NOISE
`
`____________________
`
`Inter Partes Review No. IPR2017-00626
`__________________________________________________________________
`
`Petition for Inter Partes Review of
`U.S. Patent No. 6,363,345
`
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`Table of Contents
`
`I.
`
`Introduction .................................................................................................... 1
`
`II. Regulatory Information ................................................................................ 2
`A. Certification that Petitioner May Contest the ’345 Patent
`(§ 42.104(a)) .......................................................................................... 2
`Identification of Claims Being Challenged (§ 42.104(b)) ................ 2
`Fee for Inter Partes Review (§ 42.15(a)) ............................................ 4
`
`B.
`C.
`
`III. The ’345 Patent and Background Technology ............................................ 4
`A.
`Background Technologies ................................................................... 4
`1.
`Audio Signal Processing ............................................................. 4
`2.
`Spectral Subtraction and Boll ..................................................... 7
`The ’345 Patent Disclosure ................................................................. 9
`Effective Filing Date .......................................................................... 12
`Person of Ordinary Skill in the Art ................................................. 12
`
`B.
`C.
`D.
`
`IV. Claim Construction ..................................................................................... 13
`A.
`“magnitude” ....................................................................................... 13
`B.
`“frequency spectrum generator” / “generating the frequency
`spectrum” ........................................................................................... 14
`“threshold detector for setting a threshold… and for detecting” . 15
`“generating a noise canceling signal for canceling noise” ............. 15
`
`C.
`D.
`
`V. Analysis of the Patentability of the ’345 Patent ........................................ 17
`A.
`Introduction ....................................................................................... 17
`1.
`Illustrative Claim....................................................................... 17
`2.
`Index of Grounds ...................................................................... 17
`B. Hirsch Anticipates Claims 1-3, 12-13, 21, 23, and 38 ..................... 18
`1.
`Publication ................................................................................ 18
`2.
`Overview of Hirsch ................................................................... 19
`3.
`Hirsch Anticipates Claims 1-3, 12-13, 21, 23, and 38 .............. 22
`
`
`
`ii
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`3.
`
`a)
`Independent Claim 38 ..................................................... 22
`Independent Claim 1 ....................................................... 26
`b)
`Claim 2 ............................................................................ 28
`c)
`Claim 3 ............................................................................ 29
`d)
`Claim 12 .......................................................................... 29
`a)
`Claim 13 .......................................................................... 30
`b)
`Claim 21 .......................................................................... 31
`c)
`Claim 23 .......................................................................... 31
`d)
`C. Hirsch and Martin Render Claims 4-11, 25, 39-42, and 46
`Obvious ............................................................................................... 31
`1.
`Overview of Martin ................................................................... 32
`2.
`A Skilled Person Would Have Modified the Hirsch Scheme to
`Incorporate Features Shown in Martin ..................................... 34
`Hirsch and Martin Render Claims 4-11, 25, 39-42, and 46
`Obvious ..................................................................................... 38
`a)
`Claims 4-11 and 39-41 ................................................... 38
`b)
`Claims 25 and 46 ............................................................ 48
`c)
`Claim 42 .......................................................................... 49
`D. Hirsch and Boll Render Claims 13-14, 17-21, 23, and 47 Obvious49
`1.
`Overview of Boll ....................................................................... 49
`2.
`A Skilled Person Would Have Considered Hirsch with Boll ... 51
`3.
`Hirsch and Boll Render Claims 13-14, 17-21, 23, and 47
`Obvious ..................................................................................... 53
`a)
`Claims 13-14 ................................................................... 53
`b)
`Claims 17 and 47 ............................................................ 54
`c)
`Claim 18 .......................................................................... 55
`d)
`Claim 19 .......................................................................... 56
`e)
`Claim 20 .......................................................................... 56
`f)
`Claims 21 and 23 ............................................................ 58
`E. Hirsch, Martin, and Boll Render Claim 43 Obvious ..................... 59
`
`
`
`iii
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`3.
`
`1.
`Combining Hirsch, Martin, and Boll ........................................ 59
`Hirsch, Martin, and Boll Render Claim 43 Obvious ................ 59
`2.
`F. Hirsch, Boll, and Arslan Render Claims 15-16 and 24 Obvious .. 60
`1.
`Overview of Arslan (Ex. 1011) ................................................. 60
`2.
`A Skilled Person Would Have Considered Hirsch, Boll, and
`Arslan Together ......................................................................... 61
`Hirsch, Boll, and Arslan Render Claims 15-16 and 24 Obvious62
`a)
`Claim 15 .......................................................................... 62
`b)
`Claim 16 .......................................................................... 63
`c)
`Claim 24 .......................................................................... 64
`G. Hirsch and Uesugi Render Claim 22 Obvious ................................ 66
`1.
`Overview of Uesugi (Ex. 1015) ................................................ 66
`2.
`Hirsch and Uesugi Render Claim 22 Obvious .......................... 67
`H. Hirsch, Martin, and Uesugi Render Claims 44-45 Obvious .......... 68
`1.
`Claim 44 .................................................................................... 68
`2.
`Claim 45 .................................................................................... 69
`No Secondary Considerations Exist ................................................ 70
`
`I.
`
`VI. Conclusion .................................................................................................... 70
`
`
`
`
`
`iv
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`TABLE OF AUTHORITIES
`
`Cases
`
`Page(s)
`
`Cuozzo Speed Techs., LLC v. Lee,
`136 S. Ct. 2131 (2016) ........................................................................................ 13
`
`Ericsson, Inc. v. Intellectual Ventures I LLC,
`IPR2014-00527, Paper 41 (May 18, 2015) ............................................ 19, 32, 50
`
`LG Elecs., Inc. v. Straight Path IP Grp., Inc.,
`IPR2015-00196, Paper 20 (May 15, 2015) .......................................................... 2
`
`Vivid Techs., Inc. v. Am. Sci. & Eng’g, Inc.,
`200 F.3d 795 (Fed. Cir. 1999) ............................................................................ 13
`
`Statutes
`
`35 U.S.C. § 102(b) ............................................................................................passim
`
`Other Authorities
`
`37 C.F.R. § 42.104(b) ................................................................................................ 3
`
`37 CFR § 42.100(b) ................................................................................................. 13
`
`
`
`v
`
`
`
`
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`Exhibit List
`
`1006
`
`1007
`
`1008
`
`1009
`
`Exhibit # Reference Name
`1001
`U.S. Patent No. 6,363,345
`1002
`U.S. Patent No. 6,363,345 File History
`1003
`Declaration of Bertrand Hochwald
`1004
`[Reserved]
`1005
`H. G. Hirsch and C. Ehricher, “Noise estimation techniques for
`robust speech recognition,” Proc. IEEE Int. Conf. Acoustics,
`Speech, Signal Processing, vol. 1, pp. 153 -156, 1995 (“Hirsch”)
`Rainer Martin, “An Efficient Algorithm to Estimate the
`Instantaneous SNR of Speech Signals,” Proc. Eurospeech, pp.
`1093-96, 1993 (“Martin”)
`Letter from Technische Informationsbibliothek re: Proc.
`Eurospeech 1993 (2 Jan. 2017)
`Proc. Eurospeech 1993 Vol. 2 Table of Contents from Technische
`Informationsbibliothek
`Steven F. Boll, “Suppression of Acoustic Noise in Speech Using
`Spectral Subtraction,” IEEE Transactions on Acoustics, Speech,
`and Signal Processing, Vol. ASSP-27, No. 2, April 1979 (“Boll”)
`U.S. Patent No. 5,550,924 to Helf (“Helf”)
`U.S. Patent No. 5,706,395 to Arslan (“Arslan”)
`Excerpts from Deller et al., Discrete-Time Processing of Speech
`Signals (1993)
`Excerpt from Merriam-Webster Dictionary (1993)
`Excerpts from Oppenheim and Willsky, Signals and Systems
`(1997)
`U.S. Patent No. 5,459,683 to Uesugi
`Lim and Oppenheim, “Enhancement and Bandwidth Compression
`of Noisy Speech,” Proceedings of the IEEE, vol. 67, no. 12, pp.
`1586-1604, December 1979
`Affidavit of Service in Andrea Elecs. v. Apple Inc., EDNY
`In the Matter of Certain Audio Processing Hardware and
`Software and Products Containing the Same, Inv. No. 337-TA-
`
`1010
`1011
`1012
`
`1013
`1014
`
`1015
`1016
`
`1017
`1018
`
`
`
`vi
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`1019
`
`1020
`
`Exhibit # Reference Name
`949, Claim Construction Order (U.S.I.T.C. Jan. 27, 2016) (“949
`CC Order”)
`In the Matter of Certain Audio Processing Hardware and
`Software and Products Containing Same, Inv. No. 337-TA-949,
`Complainant Andrea Electronics Corp.’s Initial Claim
`Construction Brief (U.S.I.T.C. Oct. 19, 2015) (“Andrea CC Br.”)
`In the Matter of Certain Audio Processing Hardware and
`Software and Products Containing Same, Inv. No. 337-TA-949,
`Commission Investigative Staff’s Initial Markman Brief
`(U.S.I.T.C. Oct. 19, 2015) (“OUII CC Br.”)
`Letter from the parties in 337-TA-949 informing ALJ they agreed
`to certain constructions (Nov. 10, 2015) (prior litigation)
`In the Matter of Certain Audio Processing Hardware, Software,
`and Products Containing The Same, Inv. No. 337-TA-1026,
`Verified Complaint Against Apple Inc. and Samsung Inc. Under
`Section 337 of the Tariff Act of 1930, as Amended (U.S.I.T.C.
`Sept. 19, 2016
`
`
`
`vii
`
`1021
`
`1022
`
`
`
`
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`Petitioner’s Mandatory Notices
`
`A. Real Party in Interest (§42.8(b)(1))
`
`The real party in interest of this petition pursuant to § 42.8(b)(1) is Apple
`
`Inc. (“Apple”) located at One Infinite Loop, Cupertino, CA 95014.
`
`B. Other Proceedings (§42.8(b)(2))
`
`1.
`
`Patents and Applications
`
`U.S. Patent No. 6,363,345 (“’345 patent”) is not related to any issued patents
`
`or currently pending applications. U.S. Application No. 09/385,996 claimed
`
`priority to the application that became the ’345 patent, but was abandoned.
`
`2.
`
`Related Litigation
`
`The ’345 patent has been asserted in the following litigations:
`
`• Andrea v. Apple Inc., Action No. 2-16-cv-05220 (pending);
`
`• Andrea v. Samsung Elec. Co., Action No. 2-16-cv-05217 (pending);
`
`• Andrea v. Hewlett-Packard Co., Action No. 2-15-cv-00208 (terminated);
`
`• Andrea v. Dell Inc., Action No. 2-15-00209 (terminated);
`
`• Andrea v. Acer Inc., Action No. 2-15-cv-00210 (terminated);
`
`• Andrea v. Toshiba Corp., Action No. 2-15-cv-00211 (terminated);
`
`• Andrea v. Lenovo Holding Co., Inc., Action No. 2-15-cv-00212
`
`(terminated);
`
`
`
`viii
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`• Andrea v. ASUSTeK Computer Inc., Action No. 2-15-cv-00214
`
`(terminated); and
`
`• Andrea v. Realtek Semiconductor Corp., Action No. 2-15-cv-00215
`
`(terminated).
`
`The ’345 patent has been asserted against Petitioner Apple Inc. in the
`
`following administrative proceeding before the ITC:
`
`• 337-TA-1026 (Apple Inc. and Samsung Inc., respondents) (pending).
`
`The ’345 patent has been asserted against other entities in the following
`
`administrative proceedings before the ITC:
`
`• 337-TA-949 (Waves Audio; ASUS Computer Int’l Inc.; Acer Am. Corp.;
`
`Acer Inc.; Dell Inc.; Hewlett-Packard Co.; Lenovo (United States) Inc.;
`
`Lenovo Group Ltd.; Lenovo Holding Co., Inc.; Realtek Semiconductor
`
`Corp.; Toshiba Am. Info. Sys., Inc.; Toshiba Am., Inc.; and Toshiba
`
`Corp., respondents) (terminated); and
`
`• 337-TA-3053 (ASUS Computer Int’l Inc.; ASUSTeK Computer Inc.;
`
`Acer Am. Corp.; Acer Inc.; Dell Inc.; Hewlett-Packard Co.; Lenovo
`
`(United States) Inc.; Lenovo Group Ltd.; Lenovo Holding Co., Inc.;
`
`Realtek Semiconductor Corp.; Toshiba Am. Info. Sys., Inc.; Toshiba
`
`Am., Inc.; and Toshiba Corp., respondents) (terminated).
`
`
`
`ix
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`3.
`
`Patent Office Proceedings
`
`The ’345 patent is the subject of IPR2017-00627, filed by Apple
`
`concurrently with this petition.
`
`The ’345 patent was the subject of the following proceedings before the
`
`Office:
`
`• Realtek Semiconductor Corp. v. Andrea Elecs. Corp, IPR2015-01394;
`
`• Realtek Semiconductor Corp. v. Andrea Elecs. Corp, IPR2015-01395;
`
`• Waves Audio, Ltd. v. Andrea Elecs. Corp, IPR2016-00459.
`
`C.
`
`Lead and Backup Lead Counsel (§42.8(b)(3))
`
`Lead Counsel is: Jeffrey P. Kushan (Reg. No. 43,401), jkushan@sidley.com,
`
`(202) 736-8914. Back-Up Lead Counsel are: Steven S. Baik (Reg. No. 42,281),
`
`sbaik@sidley.com, (650) 565-7016; and Thomas A. Broughan III (Reg. No.
`
`66,001), tbroughan@sidley.com, (202) 736-8314.
`
`D.
`
`Service Information (§42.8(b)(4))
`
`Service on Petitioner may be made by e-mail (iprnotices@sidley.com), mail
`
`or hand delivery to: Sidley Austin LLP, 1501 K Street, N.W., Washington, D.C.
`
`20005. The fax number for lead and backup lead counsel is (202) 736-8711.
`
`
`
`
`
`x
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`I.
`
`Introduction
`
`The ’345 patent is generally directed to methods and systems for reducing or
`
`removing noise from an audio signal. It uses a well-known and widely used signal
`
`processing technique called “spectral subtraction” that was first described in a
`
`paper by Steven Boll in 1979. Ex. 1001, 1:58-2:1 (citing Ex. 1009 (Boll)). At a
`
`high level, spectral subtraction works by taking a noisy audio signal, separating the
`
`signal into its different frequency components, estimating the noise present in each
`
`frequency component, subtracting the estimated noise value from the magnitude of
`
`frequency component, and then recombining the various frequency components
`
`back into a single signal.
`
`The ’345 patent portrays itself as an improvement to the spectral subtraction
`
`technique described in Boll. But in the two decades between Boll’s publication in
`
`1979 and the 1999 effective filing date of the ’345 patent, many others had
`
`improved Boll’s spectral subtraction technique, including by adding the
`
`functionality that the ’345 patent portrays as inventive. Most notably, Hirsch
`
`describes a spectral subtraction technique that uses an “adaptive threshold” for
`
`each frequency component to estimate the noise level. The scheme in Hirsch is
`
`indistinguishable from the independent claims, as well as many of the dependent
`
`claims, of the ’345 patent. And while Hirsch does not disclose every technique
`
`specified in the challenged dependent claims of the ’345 patent, none of those
`
`
`
`1
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`claims represents a patentable distinction from the Hirsch scheme when Hirsch is
`
`considered with other prior art. Accordingly, the Board should institute trial and
`
`cancel the challenged claims of the ’345 patent.
`
`II. Regulatory Information
`A. Certification that Petitioner May Contest the ’345 Patent
`(§ 42.104(a))
`
`Petitioner certifies that the ’345 patent is available for inter partes review.
`
`Petitioner also certifies it is not barred or estopped from requesting inter partes
`
`review of the claims of the ’345 patent. Neither Petitioner, nor any party in privity
`
`with Petitioner, has filed a civil action challenging the validity of any claim of the
`
`’345 patent. The ’345 patent has not been the subject of a prior inter partes review
`
`by Petitioner or a privy of Petitioner.
`
`Petitioner also certifies this petition for inter partes review is timely filed as
`
`this petition was filed less than one year after November 1, 2016, the date
`
`Petitioner was first served with a complaint alleging infringement of a claim of the
`
`’345 patent. See 35 U.S.C. § 315(b); Ex. 1017. Petitioner was served with an ITC
`
`complaint no earlier than September 19, 2016, (Ex. 1022), but administrative
`
`complaints do not start the one year period of § 315(b). LG Elecs., Inc. v. Straight
`
`Path IP Grp., Inc., IPR2015-00196, Paper 20 at 7-9 (May 15, 2015).
`
`Identification of Claims Being Challenged (§ 42.104(b))
`
`B.
`Petitioner challenges claims 1-25 and 38-47 based on the following grounds.
`
`
`
`2
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`(i) Claims 1-3, 12-13, 21, 23, and 38 are anticipated under 35 U.S.C.
`
`§ 102 based on H. G. Hirsch and C. Ehrlicher, “Noise estimation techniques for
`
`robust speech recognition,” Proc. IEEE Int. Conf. Acoustics, Speech, Signal
`
`Processing, vol. 1, pp. 153-156, 1995 (“Hirsch”) (Ex. 1005).
`
`(ii) Claims 4-11, 25, 39-42, and 46 are obvious under 35 U.S.C. § 103
`
`based on Hirsch and Rainer Martin, “An Efficient Algorithm to Estimate the
`
`Instantaneous SNR of Speech Signals,” Proc. Eurospeech, 1093-96, 1993
`
`(“Martin”) (Ex. 1006).
`
`(iii) Claims 13-14, 17-23, and 47 are obvious under § 103 based on
`
`Hirsch and Steven F. Boll, “Suppression of Acoustic Noise in Speech Using
`
`Spectral Subtraction,” IEEE Transactions on Acoustics, Speech, and Signal
`
`Processing, Vol. ASSP-27, No. 2, 113-20, April 1979 (“Boll”) (Ex. 1009).
`
`(iv) Claim 43 is obvious under § 103 based on Hirsch, Martin, and Boll.
`
`(v) Claims 15-16 and 24 are obvious under § 103 based on Hirsch, Boll,
`
`and U.S. Patent No. 5,706,395 to Arslan (“Arslan”) (Ex. 1011).
`
`(vi) Claim 22 is obvious under § 103 based on Hirsch and U.S. Patent No.
`
`5,459,683 to Uesugi (“Uesugi”) (Ex. 1015).
`
`(vii) Claims 44-45 are obvious under § 103 based on Hirsch, Martin, and
`
`Uesgui.
`
`
`
`3
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`Fee for Inter Partes Review (§ 42.15(a))
`
`C.
`The Director is authorized to charge the fee specified by 37 CFR § 42.15(a)
`
`to Deposit Account No. 50-1597.
`
`III. The ’345 Patent and Background Technology
`The ’345 patent specification describes a noise suppression scheme that uses
`
`the well-known spectral subtraction process described in Boll. See Ex. 1001, 1:58-
`
`2:1 (citing Ex. 1009 (Boll)). Most of the specification is directed to describing
`
`conventional features of the spectral subtraction technique or conventional digital
`
`signal processing steps, such as using a Fourier transform to convert the signal
`
`between the time domain and the frequency domain. E.g., Ex. 1001, 1:58-2:44,
`
`4:50-5:34, Fig. 1. A brief background of audio signal processing and Boll follows.
`
`A. Background Technologies
`1.
`Audio Signal Processing
`A standard graphical depiction of sound, shown below, is as a sinusoidal
`
`wave where its amplitude corresponds to how loud it is and its frequency
`
`corresponds to its pitch. Ex. 1003, ¶¶39-42.
`
`
`
`4
`
`
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`Id., ¶43. The graphs above are in the “time domain”: they depict time on the x-
`
`axis, and signal amplitude on the y-axis. Id., ¶40. The left graph is an analog
`
`signal; the right graph is a digital version of the same signal, and each blue dot is
`
`referred to as a “sample” or a “point.” Id., ¶43.
`
`Many sounds (e.g., music or human speech) are comprised of a combination
`
`of several different frequencies (e.g., a musical chord is several notes played at the
`
`same time). Id., ¶44. A graph of a sound containing multiple frequencies looks
`
`like a single signal (see middle graph below).
`
`
`
`Ex. 1012, 123, 135; Ex. 1003, ¶51. Simply inspecting the time domain
`
`representation of the signal (middle) does not reveal which frequencies are present.
`
`However, using a well-known mathematical formula called a “Fourier transform”
`
`(e.g., an “FFT”), the signal can be separated into its individual frequency
`
`components. Ex. 1003, ¶¶45-46, 50. The result is the “frequency domain”
`
`representation of the signal (right). Id., ¶46. In the figure above, the primary
`
`frequency components of the signal on the top graph are 900 Hz, 2.5 kHz, and 3.5
`
`
`
`5
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`kHz, while in the bottom signal, the primary frequency component is a 1 kHz
`
`signal. Id., ¶51.
`
`In the frequency domain, a graph of a digital audio signal typically is
`
`depicted as a histogram. Id., ¶46. The signal is divided into “frequency bins”
`
`where each bin corresponds to one of the frequencies present in the signal. Id.,
`
`¶46. The value of each frequency bin is a “complex number,” which is a pair of
`
`numbers that reflect several properties of the signal at that frequency, including its
`
`magnitude and phase. Id., ¶¶47-49. For simplicity, only the magnitude is depicted
`
`in the graph below.
`
`
`
`Processing the audio signal in the frequency domain allows certain
`
`operations to be more easily performed on each frequency individually. Ex. 1003,
`
`¶¶54-55. However, the signal must be converted back to the time domain using an
`
`inverse Fourier transform (e.g., an IFFT) before the signal can be used to recreate
`
`the sound. Id., ¶56.
`
`
`
`6
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`Dr. Hochwald provides a more detailed explanation of these basic principles
`
`and techniques of digital signal processing at ¶¶39-57 of his declaration (Ex.
`
`1003). That explanation is provided as an aid to the Board, but since those basic
`
`principles would be well known to a skilled person, they are not necessary to
`
`support the specific reasons why the claims are unpatentable in this proceeding.
`
`Spectral Subtraction and Boll
`
`2.
`Spectral subtraction is a process for removing background noise in an audio
`
`signal (e.g., the sound of a fan) that works by subtracting the noise out of the signal
`
`on a frequency by frequency (i.e., spectral) basis. Ex. 1003, ¶58.
`
`The conventional implementation of this technique is described in Boll. In
`
`it, a window function is first applied to the time-domain audio signal to divide it
`
`into a series of overlapping frames (e.g., into groups of 256 samples which might
`
`represent 50 µs of sound), and then each frame is converted to the frequency
`
`domain using an FFT. Ex. 1009, 116; Ex. 1003, ¶¶68-71.
`
`Next, an algorithm estimates the level of background noise in each
`
`frequency bin. Ex. 1009, 114. Boll calculates the noise estimate in the signal by
`
`averaging over time the values of each bin during periods where there is no speech.
`
`Id., 114, 116. To distinguish between speech and non-speech activity, Boll uses a
`
`voice switch that detects whether each audio frame as a whole contains speech.
`
`Id., 116. After the noise spectrum is estimated, it is removed from each frequency
`
`
`
`7
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`in the audio signal by subtracting the noise value (shown in gold) from the
`
`corresponding frequency bin of the signal (shown in blue). Ex. 1003, ¶¶60-63, 72-
`
`74. The result is a cleaner signal with some of the noise removed. Id., ¶62. Boll
`
`performs the subtraction step using a filter, which multiples the signal magnitude
`
`by a filter value (e.g., by 0.8, which subtracts 20% from the signal). Ex. 1009, 116.
`
`
`
`Finally, after the noise has been removed, the signal can be converted back
`
`into the time domain using an IFFT. Ex. 1009, 117. Boll explains that before
`
`converting the signal back to the time domain, additional operations can be
`
`performed on the signal to further reduce the noise. Id. For example, Boll
`
`describes two residual noise reduction processes that can further attenuate any
`
`noise remaining. Id. Boll also describes a magnitude averaging process that can
`
`be used to smooth the signal estimates over time. Id., 114.
`
`
`
`8
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`The ’345 Patent Disclosure
`
`B.
`The ’345 patent describes a noise suppression scheme based on Boll’s
`
`spectral subtraction technique. In it, a digital audio signal is first divided into
`
`frames and an FFT converts each frame from the time domain to the frequency
`
`domain. Ex. 1001, 4:65-5:14. A noise processing step then estimates the noise
`
`level in each frequency bin, and subtracts it from the signal. Id., 5:58-65. The
`
`signal is then converted back to the time domain. This process is generally
`
`illustrated in Figure 1 (below).
`
`The noise estimate for each bin is calculated by averaging over time the bin
`
`values that contain only noise. Id., 6:50-55. The ’345 patent describes use of an
`
`“exponential” averaging process that calculates the average as 0.95 times the
`
`
`
`previous estimate plus 0.05 time the new noise value:
`
`N(n) = 0.05*Y(n) + 0.95*N(n).
`
`
`
`9
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`Id., 6:53-55, Fig. 3. To determine whether a bin contains speech or only noise, the
`
`bin’s magnitude is compared to an adaptive threshold (308). Id., 6:10-19, 6:46-57.
`
`If the magnitude is less than the threshold, it is considered to be noise and is used
`
`by an exponential averaging unit (310) to calculate the noise estimate (312) for the
`
`frequency bin. Id., 6:46-55. If it is greater than the threshold, it is identified as
`
`speech and the noise estimate is not updated. Id. This process is illustrated in
`
`Figure 3 (below).
`
`
`
`The ’345 patent asserts that its method of using a separate threshold detector for
`
`each frequency bin solves a problem in the prior art—the need for a “voice switch”
`
`that detects non-speech or noise-only frames. Id., 1:58-64, 2:45-47, 3:24-31.
`
`The ’345 patent indicates that the noise estimate is used in the subtracting
`
`step to remove the noise from the signal. Id., 6:58-61. It also explains that noise
`
`can be removed from a signal using any of a number of different, but well-known
`
`
`
`10
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`mathematical operations, including simple subtraction, a filter multiplication, and a
`
`Wiener filter function. Id., 3:60-67. After the noise values have been subtracted,
`
`the signal is converted back to the time domain using an inverse FFT. Id., 9:1-5.
`
`The ’345 patent also describes a specific algorithm for setting the threshold
`
`in each frequency bin. Id., 9:54-10:18, 12:24-39 (claims 4-11, 39-41). In a
`
`preferred embodiment, the threshold is calculated by tracking the minimum value
`
`of the bin over a five second period, and then multiplying the minimum observed
`
`value by a factor of 4. Id., 6:46-48. The bin’s minimum value is tracked using two
`
`values: a “future minimum” (304), which tracks the minimum observed value
`
`during the current five second period, and a “current minimum” (306), which is
`
`used to calculate the noise threshold and reflects the minimum value observed
`
`during the previous five second period. Id., 6:10-41, Fig. 3.
`
`Other dependent claims relate to additional features of the noise reduction
`
`scheme. For example, the ’345 specification describes mathematical shortcuts that
`
`can be used to derive the magnitude of each frequency bin using the output of the
`
`FFT operation, which is a complex number that includes a real and imaginary part.
`
`Id., 3:46-51, 6:61-7:17. It also describes “smoothing” (i.e., averaging) various
`
`components of the signal by calculating an average over time or an average over
`
`neighboring frequency bins. Id., 3:21-23, 3:51-57. Finally, the ’345 patent
`
`
`
`11
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`describes a residual noise reduction process that can be used to reduce noise in the
`
`signal after spectral subtraction has been performed. Id., 4:5-11.
`
`C. Effective Filing Date
`The ’345 patent issued from U.S. Application No. 09/252,874, filed on
`
`February 18, 1999, and does not claim priority to any other application.
`
`Accordingly, the filing date of the ’345 patent claims is February 18, 1999.
`
`Person of Ordinary Skill in the Art
`
`D.
`A person of ordinary skill in the art in February 1999 would have been a
`
`person with a good working knowledge of digital signal processing techniques and
`
`their applications. The person would have gained this knowledge through an
`
`undergraduate education in electrical engineering or a comparable field, in
`
`combination with either a graduate degree (or two years of graduate work) in
`
`electrical engineering or a comparable field, or through two years of practical work
`
`experience, where such graduate education or work experience focused on or
`
`involved the use of digital signal processing techniques. Ex. 1003, ¶37.
`
`Petitioner’s positions regarding how a person of ordinary skill would have
`
`understood the ’345 patent claims and the teachings of the prior art references are
`
`supported by the testimony of Bertrand Hochwald, Ph. D., an expert in digital
`
`signal processing who has over 20 years of experience in the field. Id., ¶¶1-11.
`
`
`
`12
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`IV. Claim Construction
`Claims must be given their broadest reasonable construction in light of the
`
`specification. 37 CFR § 42.100(b); Cuozzo Speed Techs., LLC v. Lee, 136 S. Ct.
`
`2131 (2016). In this proceeding, the teachings of the asserted prior art references
`
`are squarely within the scope of the challenged claims, and consequently, the
`
`Board likely will not need to adopt specific constructions. See Vivid Techs., Inc. v.
`
`Am. Sci. & Eng’g, Inc., 200 F.3d 795, 803 (Fed. Cir. 1999) (claim terms need only
`
`be construed to the extent necessary to resolve the case).
`
`In anticipation of constructions that Patent Owner may propose, Petitioner
`
`sets forth constructions for several terms of the ’345 patent consistent with their
`
`broadest reasonable interpretation below. Petitioner reserves its right to advance
`
`different constructions in district court or ITC litigation based on a different claim
`
`construction standard, and to establish the claims are invalid under 35 U.S.C. § 112
`
`in such proceedings.
`
`“magnitude”
`
`A.
`Independent claims 1 and 38 each specify detecting whether a “magnitude”
`
`of each frequency bin is less than that bin’s threshold. The specification explains
`
`that the magnitude of the frequency bin can correspond to its actual magnitude or
`
`to a value that approximates its magnitude. Ex. 1001, 2:11-19, 2:24-30, 5:40-41
`
`(“the signal magnitude (Y) is estimated… using an approximation formula”), 5:34-
`
`
`
`13
`
`
`
`Petition in IPR2017-00626
`
`
`
`U.S. Patent No. 6,363,345
`
`44. The specification does not require use of any particular technique for
`
`measuring or estimating the magnitude of a frequency bin. Ex. 1003, ¶¶94-95.
`
`Accordingly, the broadest reasonable interpretation of the term “magnitude”
`
`includes both the signal’s actual magnitude as well as an approximation of its
`
`magnitude. Ex. 1001, 2:11-19, 2:24-30, 5:34-44.
`
`During previous litigation, Patent Owner contended the term “magnitude”
`
`was “a measure of the level of the signal in a frequenc