`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`_________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`_________________
`
`
`APPLE INC.,
`Petitioner
`
`v.
`
`ZENTIAN LIMITED,
`Patent Owner
`_________________
`
`
`Inter Partes Review Case No. IPR2023-00037
`U.S. Patent No. 10,971,140
`
`
`
`
`
`PETITION FOR INTER PARTES REVIEW
`OF U.S. PATENT NO. 10,971,140
`
`
`
`I.
`II.
`
`TABLE OF CONTENTS
`
`INTRODUCTION ........................................................................................ 1
`SUMMARY OF THE ’140 PATENT ......................................................... 1
`A. DESCRIPTION OF THE ’140 PATENT ........................................................ 1
`B.
`SUMMARY OF UNPATENTABILITY .......................................................... 1
`C.
`PRIORITY DATE OF THE CHALLENGED CLAIMS ...................................... 5
`D.
`LEVEL OF SKILL OF A POSITA .............................................................. 5
`III. REQUIREMENTS FOR IPR UNDER 37 C.F.R. § 42.104 ...................... 5
`A. GROUNDS FOR STANDING UNDER 37 C.F.R. § 42.104(A) ...................... 5
`IDENTIFICATION OF CHALLENGED CLAIMS UNDER 37 C.F.R. §
`B.
`42.104(B) AND RELIEF REQUESTED ....................................................... 6
`CLAIM CONSTRUCTION UNDER 37 C.F.R. § 42.104(B)(3) ..................... 7
`C.
`IV. SHOWING OF ANALOGOUS PRIOR ART ........................................... 7
`A.
`JIANG ..................................................................................................... 7
`B.
`CHEN ..................................................................................................... 7
`C.
`LUCKE ................................................................................................... 8
`D.
`ROBINSON .............................................................................................. 8
`E. WRENCH ................................................................................................ 8
`V. GROUND 1: CLAIMS 1-3, 5, AND 7-8 ARE OBVIOUS OVER
`JIANG IN VIEW OF CHEN ........................................................................ 9
`A. OBVIOUSNESS OF JIANG COMBINED WITH CHEN .................................... 9
`1. Motivation to Combine Jiang and Chen .................................... 9
`Obviousness of Modifying Electronic Hardware and
`2.
`Software to Be Circuitry .......................................................... 11
`CLAIM 1 .............................................................................................. 12
`Claim 1(Pre): “A speech recognition circuit
`1.
`comprising:” ............................................................................ 12
`Claim 1(a): “one or more clusters of processors, each of
`the one or more clusters of processors comprising: a
`plurality of processors;” .......................................................... 14
`a)
`Jiang Teaches a Multiprocessor System ....................... 14
`b)
`The ’140 Patent’s Description of a “Cluster” ................ 14
`c)
`Chen Teaches Claim 1(a) .............................................. 15
`d) Motivation to Combine Chen with Jiang for Claim
`1(a) ................................................................................. 17
`
`B.
`
`2.
`
`ii
`
`
`
`3.
`
`4.
`
`b)
`
`b)
`
`Claim 1(b): “and [one or more clusters of processors,
`each of the one or more clusters of processors
`comprising:] an acoustic model memory storing acoustic
`model data” ............................................................................. 19
`Jiang Teaches an Acoustic Model Memory Storing
`a)
`Acoustic Model Data ..................................................... 19
`Chen Teaches One or More Clusters of Processors
`Each Comprising Memory ............................................. 22
`c) Motivation to Combine Chen with Jiang for Claim
`1(b) ................................................................................. 24
`Claim 1(c): “wherein each of the plurality of processors
`is configured to compute a probability using the acoustic
`model data in the acoustic model memory” ............................ 25
`Jiang Teaches “comput[ing] a probability using
`a)
`the acoustic model data” ................................................ 25
`Jiang in Combination with Chen Teaches “each of
`the plurality of processors [is configured to
`compute] … using the acoustic model data in the
`acoustic model memory” ............................................... 28
`c) Motivation to Combine Chen with Jiang for Claim
`1(c) ................................................................................. 31
`Claim 1(d): “[wherein] the speech recognition circuit is
`configured to generate an initial score for an audio
`sample” .................................................................................... 31
`a)
`Jiang Teaches an “audio sample” .................................. 31
`b)
`Jiang Teaches “the speech recognition circuit is
`configured to generate an initial score” ......................... 32
`Claim 1(e): “[wherein] the initial score is used to
`determine whether to continue processing to determine a
`final score via processing a larger amount of model data
`than that was processed to generate the initial score” ............ 36
`Prosecution History of the ’140 Patent Regarding
`a)
`Claim 1(e) ...................................................................... 36
`Jiang Teaches Claim 1(e) .............................................. 38
`b)
`CLAIM 2: ............................................................................................. 44
`Claim 2: “…wherein the probability is an input to an
`1.
`evaluation of a state transition of a model of states” .............. 44
`CLAIM 3 .............................................................................................. 47
`Claim 3: “…wherein the model is a Hidden Markov
`1.
`Model” ..................................................................................... 47
`
`5.
`
`6.
`
`C.
`
`D.
`
`iii
`
`
`
`E.
`
`F.
`
`G.
`
`CLAIM 5 .............................................................................................. 47
`Claim 5: “…further comprising: a buffer for storing one
`1.
`or more feature vectors coupled to at least one of the
`plurality of processors of the one or more clusters of
`processors” .............................................................................. 47
`CLAIM 7 .............................................................................................. 49
`Claim 7: “…the one or more clusters of processors
`1.
`comprises a first cluster of processors and a second
`cluster of processors; the first cluster comprises a first
`acoustic model memory; and ................................................... 49
`the second cluster comprises a second acoustic model memory
`that is distinct and separate from the first acoustic model
`memory” .................................................................................. 49
`CLAIM 8 .............................................................................................. 51
`Claim 8: “…the one or more processors comprises a first
`1.
`cluster of processors; and ........................................................ 51
`the first cluster comprises a first acoustic model coupled to all
`processors of the first cluster.” ................................................ 51
`VI. GROUND 2: CLAIMS 1-3, 5, AND 7-8 ARE OBVIOUS OVER
`JIANG IN VIEW OF CHEN IN FURTHER VIEW OF LUCKE ........... 53
`A. OVERVIEW OF GROUND 2 .................................................................... 53
`B.
`CLAIM 1 .............................................................................................. 54
`Claim 1(Pre): “A speech recognition circuit
`1.
`comprising:” ............................................................................ 54
`Claim 1(a): “one or more clusters of processors, each of
`the one or more clusters of processors comprising: a
`plurality of processors;” .......................................................... 54
`Claim 1(b): “and [one or more clusters of processors,
`each of the one or more clusters of processors
`comprising:] an acoustic model memory storing acoustic
`model data” ............................................................................. 54
`Claim 1(c): “wherein each of the plurality of processors
`is configured to compute a probability using the acoustic
`model data in the acoustic model memory, wherein:” ............ 54
`Claim 1(d): “the speech recognition circuit is configured
`to generate an initial score for an audio sample; and ............. 54
`Claim 1(e): “the initial score is used to determine
`whether to continue processing to determine a final score
`
`2.
`
`3.
`
`4.
`
`5.
`
`6.
`
`iv
`
`
`
`C.
`
`D.
`
`E.
`
`G.
`
`via processing a larger amount of model data than that
`was processed to generate the initial score” ........................... 54
`CLAIM 2 .............................................................................................. 57
`Claim 2: “…the probability is an input to an evaluation
`1.
`of a state transition of a model of states” ................................ 57
`CLAIM 3 .............................................................................................. 57
`1.
`Claim 3: “…the model is a Hidden Markov Model” ............... 57
`CLAIM 5 .............................................................................................. 57
`Claim 5: “…a buffer for storing one or more feature
`1.
`vectors coupled to at least one of the plurality of
`processors of the one or more clusters of processors” ............ 57
`CLAIM 7 .............................................................................................. 58
`Claim 7: “…the one or more clusters of processors
`1.
`comprises a first cluster of processors and a second
`cluster of processors; ............................................................... 58
`the first cluster comprises a first acoustic model memory; and ......... 58
`the second cluster comprises a second acoustic model memory
`that is distinct and separate from the first acoustic model
`memory” .................................................................................. 58
`CLAIM 8 .............................................................................................. 58
`Claim 8: “…the one or more processors comprises a first
`1.
`cluster of processors; and ........................................................ 58
`the first cluster comprises a first acoustic model coupled to all
`processors of the first cluster.” ................................................ 58
`VII. GROUND 3: CLAIM 4 IS OBVIOUS OVER JIANG IN VIEW
`OF CHEN IN FURTHER VIEW OF ROBINSON .................................. 59
`CLAIM 4: “…WHEREIN THE PROBABILITY IS COMPUTED FROM A
`A.
`GAUSSIAN MIXTURE MODEL AND ONE OR MORE FEATURE
`VECTORS” ............................................................................................ 59
`VIII. GROUND 4: CLAIM 4 IS OBVIOUS OVER JIANG IN VIEW
`OF CHEN IN FURTHER VIEW OF LUCKE IN FURTHER
`VIEW OF ROBINSON ............................................................................... 60
`A.
`CLAIM 4 .............................................................................................. 60
`IX. GROUND 5: CLAIM 6 IS OBVIOUS OVER JIANG IN VIEW
`OF CHEN IN FURTHER VIEW OF WRENCH ..................................... 60
`CLAIM 6: “…A SEARCH CONTROLLER COUPLED TO THE AT LEAST
`A.
`ONE OF THE PLURALITY OF PROCESSORS OF THE ONE OR MORE
`CLUSTERS OF PROCESSORS, THE SEARCH CONTROLLER CAPABLE
`
`F.
`
`v
`
`
`
`OF CONTROLLING THE AT LEAST ONE OF THE PLURALITY OF
`PROCESSORS OF THE ONE OR MORE CLUSTERS OF PROCESSORS TO
`INITIATE SPEECH RECOGNITION PROCESSING IN AT LEAST A
`SUBSET OF PROCESSORS OF THE ONE OR MORE CLUSTERS OF
`PROCESSORS” ...................................................................................... 60
`X. GROUND 6: CLAIM 6 IS OBVIOUS OVER JIANG IN VIEW
`OF CHEN IN FURTHER VIEW OF LUCKE IN FURTHER
`VIEW OF WRENCH .................................................................................. 62
`A.
`CLAIM 6 .............................................................................................. 62
`XI. THE BOARD’S DISCRETION UNDER 35 U.S.C. § 314(A) ................. 62
`XII. CONCLUSION ........................................................................................... 63
`XIII. MANDATORY NOTICES UNDER 37 C.F.R. § 42.8(A)(1) ................... 64
`A.
`REAL PARTY-IN-INTEREST .................................................................. 64
`B.
`RELATED MATTERS ............................................................................. 64
`C.
`LEAD AND BACK-UP COUNSEL ........................................................... 64
`FAX: (913) 777-5601 ..................................................................................... 65
`
`
`
`
`
`
`vi
`
`
`
`TABLE OF AUTHORITIES
`
`Cases:
`
`Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005)
`
`Statutes:
`35 U.S.C. § 102(b)
`35 U.S.C. § 102(e)
`35 U.S.C. § 314(a)
`
`Regulations:
`37 C.F.R. § 42.6(e)
`37 C.F.R. § 42.8
`37 C.F.R. § 42.8(a)(1)
`37 C.F.R. § 42.8(b)(1)
`37 C.F.R. § 42.8(b)(3)
`37 C.F.R. § 42.8(b)(4)
`37 C.F.R. § 42.24
`37 C.F.R. § 42.100(b)
`37 C.F.R. § 42.104
`37 C.F.R. § 42.104(a)
`37 C.F.R. § 42.104(b)
`37 C.F.R. § 42.104(b)(1)
`37 C.F.R. § 42.104(b)(2)
`37 C.F.R. § 42.104(b)(3)
`37 C.F.R. § 42.104(b)(4)
`37 C.F.R. § 42.104(b)(5)
`37 C.F.R. § 42.105
`
`
`vii
`
`7
`
`7-8
`7-8
`62
`
`72
`71
`65
`65
`65
`65
`71
`7
`5
`5
`6
`6
`6
`7
`6
`6
`72
`
`
`
`I.
`
`INTRODUCTION
`Petitioner Apple Inc. requests Inter Partes Review of Claims 1-8
`
`(collectively, the “Challenged Claims”) of USPN 10,971,140 (“’140 Patent”). (Ex.
`
`1001). The purportedly distinguishing features of the Challenged Claims—using
`
`clustered processors and determining whether to continue processing speech using a
`
`larger amount of model data based on an initial score—were well-known before the
`
`priority date of the ’140 Patent, and the Challenged Claims are obvious over the prior
`
`art as detailed herein. Accordingly, IPR of the Challenged Claims should be
`
`instituted.
`
`II.
`
`SUMMARY OF THE ’140 PATENT
`A. Description of the ’140 Patent
`The ’140 Patent is directed to a speech recognition circuit utilizing clustered
`
`processors to perform speech recognition using an acoustic model stored in memory.
`
`’140 Patent, 1:18-20, 5:6-29, Fig. 2. The claimed circuit generates an initial score
`
`for an audio sample, used to determine which possible word paths in a lexical tree
`
`should be “pruned” from its search. ’140 Patent, 6:21-35, 8:59-9:4.
`
`B.
`Summary of Unpatentability
`Claim 1’s speech recognition circuit generally recites an acoustic model
`
`memory, well-known speech recognition steps associated with generating scores to
`
`inform processing of an audio sample, and one or more clusters of processors for
`
`performing the speech recognition steps. The proposed combination for Ground 1,
`1
`
`
`
`
`
`Claim 1 relies on Jiang and Chen. Jiang is applied to teach the well-known speech
`
`recognition system comprising an acoustic model memory and speech recognition
`
`steps. Chen is applied to teach the well-known clusters of processors. Performing
`
`speech recognition using clusters of processors would have been obvious at the time
`
`of the ’140 Patent. Indeed, Jiang already teaches a multi-processor configuration,
`
`and clusters of processors was known to “create a high performance parallel
`
`processing computer system.” Jiang, 4:60-65, 6:39-42; Chen, 5:9-13.
`
`Jiang teaches a speech recognition system comprising a CPU 21, “which may
`
`include one or more processors[.]” Jiang, 6:39-42, 4:60-65. Jiang teaches utilizing
`
`“phonetic speech unit models, such as hidden Markov models, which represent
`
`speech units to be detected by system 60.” Jiang, 7:34-36, 7:36-38 (teaching the
`
`stored phonetic models “includes HMMs which represent phonemes”). A Hidden
`
`Markov Model (HMM) comprises a plurality of states for determining word or sub-
`
`word units (e.g., phonemes) that can be organized into a lexical tree. A “conventional
`
`technique” for reducing computation requirements, referred to in Jiang as a “prefix
`
`tree,” associates phonetic models, such as those modeled by HMMs, with the
`
`branches of the prefix tree. Jiang, 2:47-59; Ravishankar, 33, 67-71 (generally
`
`describing the known benefits of the lexical tree structure) (Ex. 1007); Dec., 84
`
`(opining a POSITA would have understood Jiang’s “prefix tree” to be a “lexical
`
`tree”). In Jiang, the prefix tree is representative of phonemes, and the prefix tree
`
`
`
`2
`
`
`
`comprises a plurality of nodes connected by phoneme branches. Jiang, 2:53-57. A
`
`score is assigned to each node of the prefix tree as the tree is traversed. Jiang, 8:16-
`
`24. Jiang’s speech recognition circuit computes “probability score[s]” for each state
`
`of an acoustic model stored in memory, such as the HMM, and then uses the output
`
`probabilities to search the prefix tree in a “known manner.” Jiang, 2:3-15, 1:19-32,
`
`8:16-51.
`
`Jiang, Fig. 4, 1:19-2:15, 9:28-42. Using initial generated scores, Jiang’s speech
`
`recognition system prunes branches from the search, continuing to process (using
`
`
`
`
`
`3
`
`
`
`additional data from the acoustic model) only branches whose score is sufficiently
`
`high. Jiang, 2:25-46, 8:52-64, 10:51-11:15.
`
`Jiang does not teach clusters of processors. In related art, Chen teaches a
`
`“parallel processing architecture for connecting together an extendible number of
`
`clusters of multiple numbers of processors to create a high performance parallel
`
`processing computer system.” Chen, 5:9-13. As illustrated in annotated Fig. 4 below,
`
`Chen’s clusters 100a-d each includes a “Cluster Shared Memory” 104a-d accessible
`
`to each of the plurality of processors 102a-d in the respective cluster.
`
`
`
`4
`
`
`
`
`
`Chen, Fig. 4 (annotated), 9:5-43. Thus, the speech recognition circuit of Jiang
`
`modified to utilize Chen’s architecture of a plurality of clustered processors renders
`
`obvious every limitation of the Challenged Claims, as detailed in the mapping below.
`
`C.
`Priority Date of the Challenged Claims
`The ’140 Patent claims priority to a GB application filed February 4, 2002.
`
`’140 Patent, (22), (63), (30). For purposes of this Petition, Apple applies February
`
`4, 2002, as the priority date for the Challenged Claims.
`
`D. Level of Skill of a POSITA
`A POSITA at the time of the ’140 Patent would have had a master’s degree in
`
`computer engineering, computer science, electrical engineering, or a related field,
`
`with at least two years of experience in the field of speech recognition, or a
`
`bachelor’s degree in the same fields with at least four years of experience in the field
`
`of speech recognition. Additional education or experience might substitute for the
`
`above requirements. Dec., 24-26.1
`
`III. REQUIREMENTS FOR IPR UNDER 37 C.F.R. § 42.104
`A. Grounds for Standing Under 37 C.F.R. § 42.104(a)
`Apple certifies the ’140 Patent is available for IPR and Apple is not barred or
`
`estopped from requesting IPR challenging the claims of the ’140 Patent. Apple is
`
`not the owner of the ’140 Patent, has not filed a civil action challenging the validity
`
`
`1All citations to “Dec.” are to paragraph numbers in Ex. 1003, Declaration of
`Christopher Schmandt.
`
`
`
`5
`
`
`
`of any claim of the ’140 Patent, and this Petition is filed less than one year after
`
`Apple was served with a complaint alleging infringement of the ’140 Patent.
`
`B.
`
`Identification of Challenged Claims Under 37 C.F.R. § 42.104(b)
`and Relief Requested
`In view of the prior art and evidence presented, the Challenged Claims of the
`
`’140 Patent are unpatentable and should be cancelled. 37 C.F.R. § 42.104(b)(1).
`
`Based on the prior art identified below, IPR of the Challenged Claims should be
`
`granted. 37 C.F.R. § 42.104(b)(2).
`
`Ground 1:
`
`Proposed Grounds of Unpatentability
`Claims 1-3, 5, and 7-8 are obvious over Jiang and Chen
`
`Ground 2:
`
`Claims 1-3, 5, and 7-8 are obvious over Jiang, Chen, and Lucke
`
`Ground 3:
`
`Claim 4 is obvious over Jiang, Chen, and Robinson
`
`Ground 4:
`
`Claim 4 is obvious over Jiang, Chen, Lucke, and Robinson
`
`Ground 5:
`
`Claim 6 is obvious over Jiang, Chen, and Wrench
`
`Ground 6:
`
`Claim 6 is obvious over Jiang, Chen, Lucke, and Wrench
`
`
`
`Sections V-X identify where each element of the Challenged Claims is found
`
`in the prior art. 37 C.F.R. § 42.104(b)(4). The exhibit numbers of the supporting
`
`evidence relied upon to support the challenges are provided above and the relevance
`
`of the evidence to the challenges raised are provided in Sections V-X. 37 C.F.R.
`
`§ 42.104(b)(5). Exhibits 1001–1033 are also attached.
`
`
`
`6
`
`
`
`C. Claim Construction Under 37 C.F.R. § 42.104(b)(3)
`In this proceeding, claims are interpreted under the same standard applied by
`
`Article III courts (i.e., the Phillips standard). 37 C.F.R. § 42.100(b); 83 Fed. Reg.
`
`197 (Oct. 11, 2018); Phillips v. AWH Corp., 415 F.3d 1303, 1312 (Fed. Cir. 2005)
`
`(en banc). Petitioner applies the plain and ordinary meaning of all claim terms.
`
`Petitioner does not waive any argument in any litigation that claim terms in the ’140
`
`Patent are indefinite or additional terms need construction.
`
`IV. SHOWING OF ANALOGOUS PRIOR ART
`A.
`Jiang
`Jiang was neither cited nor considered during prosecution of the ’140 Patent.
`
`Jiang was filed February 20, 1998, and issued April 16, 2002, qualifying as prior art
`
`under 35 U.S.C. § 102(e). Jiang describes a speech recognition circuit utilizing an
`
`HMM acoustic model. Compare Jiang (Ex. 1004), 6:18-28, 7:29-54, with ’140
`
`Patent, 1:30-44, 6:41-48; Dec., 62. Thus, Jiang is analogous art to the ’140 Patent.
`
`B.
`Chen
`Chen was neither cited nor considered during prosecution of the ’140 Patent.
`
`Chen issued June 27, 1995, qualifying as prior art under 35 U.S.C. § 102(b). Chen
`
`describes a parallel processing computing platform utilizing clustered processors.
`
`Compare Chen (Ex. 1005), 5:9-17, 9:5-43, Fig. 4, with ’140 Patent, 3:13-24, 5:6-36,
`
`Fig. 2; Dec., 63. Thus, Chen is analogous art to the ’140 Patent.
`
`
`
`7
`
`
`
`C.
`Lucke
`Lucke was neither cited nor considered during prosecution of the ’140 Patent.
`
`Lucke was filed March 12, 2001, and published December 20, 2001, qualifying as
`
`prior art under 35 U.S.C. § 102(e). Lucke describes a speech recognition circuit
`
`utilizing an HMM acoustic model. Compare Lucke (Ex. 1008), 0002-0005, 0008,
`
`0025, 0079 with ’140 Patent, 1:30-44, 6:41-48; Dec., 64. Thus, Lucke is analogous
`
`art to the ’140 Patent.
`
`D. Robinson
`Robinson was neither cited nor considered during prosecution of the ’140
`
`Patent. Robinson issued November 9, 1999, qualifying as prior art under 35 U.S.C.
`
`§ 102(b). Robinson describes a speech recognition circuit comparing feature vectors
`
`to a speech model. Compare Robinson (Ex. 1009), 8:28-55, 9:4-16, with ’140 Patent,
`
`4:52-67, 5:6-26; Dec., 65. Thus, Robinson is analogous art to the ’140 Patent.
`
`E. Wrench
`Wrench was neither cited nor considered during prosecution of the ’140
`
`Patent. Wrench issued July 30, 1991, qualifying as prior art under 35 U.S.C.
`
`§ 102(b). Wrench describes a speech recognition circuit comprising a plurality of
`
`processors directed by a single controller. Compare Wrench (Ex. 1010), 4:40-58,
`
`6:38-7:7, Fig. 2 with ’140 Patent, 5:46-52, 10:3-30, Fig. 2; Dec., 66. Thus, Wrench
`
`is analogous art to the ’140 Patent.
`
`
`
`8
`
`
`
`V. GROUND 1: CLAIMS 1-3, 5, AND 7-8 ARE OBVIOUS OVER JIANG
`IN VIEW OF CHEN
`A. Obviousness of Jiang Combined with Chen
`1. Motivation to Combine Jiang and Chen
`As discussed, Jiang and Chen are analogous art to the ’140 Patent, and a
`
`POSITA would have been familiar with both references. A POSITA would have
`
`found it obvious and been motivated to combine Chen’s cluster of processors with
`
`shared memory with Jiang’s speech recognition circuit using an HMM acoustic
`
`model to generate scores, as taught by Jiang. Dec., 67. Computing platforms
`
`providing processors in a cluster were extremely well-known in the art prior to the
`
`’140 Patent, as taught by Chen and numerous other references discussed by Mr.
`
`Schmandt. Dec., 67. Prior to the priority date of the ’140 Patent, connecting a
`
`plurality of processors in parallel was a well-known technique to increase the
`
`processing power available to a computing circuit without becoming expensive to
`
`produce. Dec., 67. It was also well-known that speech recognition was a
`
`computationally-intensive task, requiring expensive, sophisticated processors (for
`
`the time) if performed in real-time on a solitary, single-core processor. Dec., 67.
`
`Well before the priority date of the ’140 Patent, clusters of processors were used to
`
`recognize speech. Id.
`
`The combination of the disclosures of Jiang and Chen would have constituted
`
`use of a known technique (e.g., Chen’s clustering of processors with memory) to
`
`
`
`9
`
`
`
`improve a similar device (Jiang’s multiprocessor speech recognition circuit) in the
`
`same way. Dec., 67. A POSITA would have recognized utilizing a clustered
`
`processor architecture with shared memory would have enabled the speech
`
`recognition circuit to efficiently process branches of the lexical tree in parallel. Dec.,
`
`67. A POSITA would have understood a cluster of processors, as described in Chen,
`
`was usable and desirable in a speech recognition system with multiple processors,
`
`such as Jiang’s circuit, and would have been a low-cost, simple-to-implement
`
`improvement. Specifically, a POSITA would have recognized that the increased
`
`computing resources provided by Chen’s cluster of processors would have increased
`
`the speed of the system, enabling a reduced cost system that recognized speech at a
`
`faster rate and/or higher level of accuracy. Dec., 67.
`
`There would have been a reasonable expectation of success for adding Chen’s
`
`clusters to Jiang’s system because Chen and Jiang each contemplates a hardware
`
`architecture using multiple signal processors; therefore, the more efficient
`
`processing architecture of Chen would have been simply substituted for Jiang’s
`
`architecture to achieve a compatible improvement in processing speed and power.
`
`Dec., 67. Given the ubiquitous use of pluralities of processors in speech recognition
`
`circuits, a POSITA would have had a reasonable expectation of success, without
`
`undue experimentation, in modifying Jiang’s circuit to use Chen’s clusters of
`
`processors. Dec., 67.
`
`
`
`10
`
`
`
`Additional reasons for the combination are provided in the mapping.
`
`2.
`
`Obviousness of Modifying Electronic Hardware and Software
`to Be Circuitry
`To the extent Patent Owner argues the various elements cited from Jiang,
`
`Chen, Lucke, Robinson, or Wrench must be embodied as application-specific
`
`“circuitry,” it would have been obvious to a POSITA to implement any of the
`
`computerized techniques or physical architectures described performing similar
`
`functions to the ’140 Patent’s claims using “circuitry.” Dec., 69. A POSITA would
`
`have recognized each of the references independently describes computerized
`
`methods performed by electrical hardware components including computer
`
`processors, memory, a power source, and electrical connections therebetween, thus
`
`teaching or at least rendering obvious “circuitry” performing the recited methods.
`
`Dec., 69. Specifically, to the extent any particular description refers to computerized
`
`methods stored as software, a POSITA would have physically implemented those
`
`computerized methods using electronic circuitry. Dec., 69. Therefore, a POSITA
`
`would have found it obvious and been motivated to embody the various electronic
`
`components disclosed
`
`in
`
`the
`
`references as “circuitry” without undue
`
`experimentation. Dec., 69.
`
`
`
`11
`
`
`
`B. Claim 1
`1.
`Claim 1(Pre): “A speech recognition circuit comprising:”
`To the extent the preamble is limiting, Jiang teaches or at least renders
`
`obvious a speech recognition circuit, as claimed. Jiang, 1:15-18, 6:18-61, 4:51-5:3,
`
`Abstract, Figs. 1, 2, 7; Dec., 70. Jiang teaches a speech recognition system 60 that
`
`includes a microphone 62, A/D converter 64, and program modules 65-75.
`
`Jiang, Fig. 2, 6:18-28. Jiang specifically teaches its speech recognition system may
`
`be implemented on a suitable computing environment such as the personal computer
`
`illustrated in Fig. 1.
`
`
`
`
`
`12
`
`
`
`
`
`Jiang, Fig. 1, 4:51-5:3.
`
`Per Mr. Schmandt, a POSITA would have understood Jiang’s speech
`
`recognition system implemented on a computer teaches or at least renders obvious a
`
`“speech recognition circuit.” See Section VI.B.2; Dec., 70. Jiang already teaches
`
`“those skilled in the art will appreciate that the invention may be practiced with other
`
`computer system configurations….” Jiang, 4:60-62. Therefore, per Mr. Schmandt,
`
`given that it is generally obvious to configure computerized functions in circuity,
`
`and further given Jiang expressly teaches its invention may be employed in a variety
`
`of “computer system configurations,” it would have been obvious to a POSITA to
`
`implement Jiang’s speech recognition functionality as circuitry. Dec., 70.
`
`
`
`13
`
`
`
`2.
`
`Claim 1(a): “one or more clusters of processors, each of the
`one or more clusters of processors comprising: a plurality of
`processors;”
`a)
`Jiang Teaches a Multiprocessor System
`Jiang in view of Chen teaches or at least renders obvious this limitation. Jiang
`
`teaches its speech recognition system may be implemented on various computing
`
`systems including specifically “multiprocessor systems, microprocessor-based or
`
`programmable consumer electronics, network PCs, minicomputers, mainframe
`
`computers, and the like.” Jiang, 4:51-5:3, see also 6:39-42 (“…CPU 21 [] may
`
`include one or more processors…”). However, Jiang does not specifically discuss
`
`clusters of processors. As discussed below in Section V.B.2.c, Chen teaches clusters
`
`of processors, with each cluster comprising a plurality of processors and memory.
`
`Chen, 9:5-43.
`
`b)
`The ’140 Patent’s Description of a “Cluster”
`The ’140 Patent describes a “cluster” as at least including processing
`
`architectures comprising a group of processors and a memory. ’140 Patent, 3:14-19
`
`(“Thus a group of lexical tree processors and a partial lexical memory form a
`
`cluster.”), 3:53-57, 5:14-21, Fig. 2. Computing systems arranged in a clustered
`
`architecture were widely known and broadly used for a wide array of computing
`
`applications, including specifically speech recognition, prior to the ’140 Patent.
`
`Dec., 72 (citing examples of clustered speech recognition systems described by Hon,
`
`Tremblay, Erman, Wulf, Garde, and Deo). As discussed by Hon, computing
`14
`
`
`
`
`
`platforms performing speech recognition, such as the MARS multiprocessor
`
`machine, benefitted from an architecture comprising a plurality of “clusters” of
`
`processing elements, providing desirably improved scalability and flexibility to
`
`speech recognition applications. Hon, pp. 24-25 (Ex. 1006); Dec., 72 (further noting
`
`the link between Jiang’s “multiprocessor” system and Hon’s description of the
`
`MARS “clustered” system as “multiprocessor”).
`
`c)
`Chen Teaches Claim 1(a)
`An example of a clustered computing platform available in the art prior to the
`
`’140 Patent is described by Chen, teaching a “parallel processing computer system”
`
`comprising a plurality of “clusters” of processors. Chen, 9:5-19. As seen below in
`
`annotated Fig. 4, each cluster 100a-d includes two or more processors 102a-d “that
`
`are symmetrically connected to a cluster shared memory 104a via a connection node
`
`106a.”
`
`
`
`15
`
`
`
`
`
`Chen, Fig. 4 (annotated), 9:5-19; Dec., 73.
`
`The resulting computing architecture is described by Chen as “an extendible
`
`number of clusters of multiple numbers of processors to create a high
`
`performance parallel processing computer system.” Chen, 5:9-17. Thus, Chen
`
`teaches the claimed “one or more clusters of processors,” namely Chen’s clusters
`
`100a, 100b, 100c, and 100d. Chen, 9:13-14, Fig. 4. Chen further teaches the claimed
`
`
`
`16
`
`
`
`“each of the one or more clusters of processors comprising: a plurality of
`
`processors[,]” namely clusters 100a-d each comprising its respective processors
`
`102a, 102b, 102c, and 102d. Chen, Abstract, 5:9-17, 6:17-21, 9:5-43, 10:14-35
`
`(envisioning that Chen’s clustered configuration may be formed using commercially
`
`available microprocessors), Figs. 4, 5A, 6A-B; Dec., 73.
`
`d) Motivation to Combine Chen with Jiang for Claim 1(a)
`A POSITA would have been motivated