`571-272-7822
`
`Paper 10
`Entered: June 12, 2023
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`APPLE INC.,
`Petitioner,
`v.
`ZENTIAN LIMITED,
`Patent Owner.
`
`IPR2023-00037
`Patent 10,971,140 B2
`
`
`
`
`
`
`
`
`
`Before KEVIN F. TURNER, JEFFREY S. SMITH, and
`CHRISTOPHER L. OGDEN, Administrative Patent Judges.
`SMITH, Administrative Patent Judge.
`
`DECISION
`Granting Institution of Inter Partes Review
`35 U.S.C. § 314
`
`
`
`
`
`
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`
`INTRODUCTION
`I.
`A. Background and Summary
`Petitioner filed a Petition (Paper 1, “Pet.”) requesting inter partes
`review of claims 1–8 of U.S. Patent No. 10,971,140 B2 (Ex. 1001, “the
`’140 patent”) pursuant to 35 U.S.C. § 311(a). Patent Owner filed a
`Preliminary Response (Paper 6, “Prelim. Resp.”) pursuant to 35 U.S.C.
`§ 313.
`Pursuant to 35 U.S.C. § 314(a), the Director may not authorize an
`inter partes review unless the information in the petition and preliminary
`response “shows that there is a reasonable likelihood that the petitioner
`would prevail with respect to at least 1 of the claims challenged in the
`petition.” For the reasons that follow, we institute an inter partes review as
`to claims 1–8 of the ’140 patent on all grounds of unpatentability asserted in
`the Petition.
`
`B. Related Matters
`The parties indicate that the following matters relate to the ’140
`patent: Zentian Ltd v. Apple Inc., 6:22-cv-00122, (W.D.Tex. Feb. 2, 2022);
`Zentian Ltd v. Amazon.com, Inc., 6:22-cv-00123, (W.D.Tex. Feb. 2, 2022);
`Apple Inc. v. Zentian Ltd., Inter Partes Review No. IPR2023-00033; Apple
`Inc. v. Zentian Ltd., Inter Partes Review No. IPR2023-00034; Apple Inc. v.
`Zentian Ltd., Inter Partes Review No. IPR2023-00035; and Apple Inc. v.
`Zentian Ltd., Inter Partes Review No. IPR2023-00036. Paper 4, 1; Pet. 64.
`C. The ’140 Patent
`The ’140 patent is related to a speech recognition circuit which uses
`parallel processors for processing the input speech data in parallel.
`Ex. 1001, 1:18–20.
`
`2
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`The patent describes that in speech recognition, there are generally
`two processes: “front end processing to generate processed speech
`parameters such as feature vectors, followed by a search process which
`attempts to find the most likely set of words spoken from a given vocabulary
`(lexicon).” Id. at 1:21–26. According to the ’140 patent, “for large
`vocabulary, speaker independent speech recognition, it is the search process
`that presents the biggest challenge.” Id. at 1:28–30.
`The ’140 describes that in order to speed up the search function,
`parallel processing techniques have been suggested. Id. at 1:45–47. The
`patent further describes that “one algorithm for performing the search is the
`Viterbi algorithm,” which “is a parallel or breadth first search through a
`transition network of states of Hidden Markov Models.” Id. at 1:36–39.
`This search algorithm is computationally intensive, but in one paper cited by
`the ’140 patent, “a multi-threaded implementation of a fast beam search
`algorithm is disclosed.” Id. at 1:47–52. This “multi-threading
`implementation requires a significant amount of communication and
`synchronization among threads,” but in another paper, “the parallel
`processing of input speech parameters is disclosed in which a lexical
`network is split statically among processors.” Id. at 1:52–58.
`To implement parallel processing of the search function, the ’140
`describes a special circuit, in which a “plurality of lexical tree processors are
`connected in parallel to the input port and perform parallel lexical tree
`processing for word recognition by accessing the lexical data in the lexical
`memory arrangement.” Id. at 2:4–8. In addition, a “controller controls the
`lexical tree processors to process lexical trees identified in the results
`memory arrangement by performing parallel processing of a plurality of
`lexical tree data structures.” Id. at 2:12–15.
`
`3
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`Figure 2 is a diagram of the circuit of the ’140 patent, and is
`reproduced below.
`
`
`Figure 2, showing a plurality k of lexical tree processors 21, arranged
`in a lexical tree processor cluster 22, with acoustic model memory 23.
`D. Illustrative Claim
`Challenged claim 1 of the ’140 patent recites:
`1. [Pre] A speech recognition circuit comprising:
`[a] one or more clusters of processors, each of the one or more
`clusters of processors comprising:
`a plurality of processors; [b] and
`an acoustic model memory storing acoustic model data, [c]
`wherein each of the plurality of processors is
`configured to compute a probability using the
`acoustic model data in the acoustic model memory,
`[d] wherein:
`the speech recognition circuit is configured to
`generate an initial score for an audio sample;
`[e] and
`
`4
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`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.
`Ex. 1001, 6:13–26; Pet. 66 (showing Petitioner’s bracketed claim
`annotations).
`
`E. Evidence
`Petitioner relies on the following prior art:
`US Patent No. 6,374,219 B1, issued April 16, 2002 (Ex. 1004,
`“Jiang”);
`US Patent No. 5,428,803, issued June 27, 1995 (Ex. 1005,
`“Chen”);
`US Patent Appl. Publ. No. 2001/0053974 A1, published
`December 20, 2001 (Ex. 1008, “Lucke”);
`US Patent No. 5,983,180, issued November 9, 1999 (Ex. 1009,
`“Robinson”);
`US Patent No. 5,036,539, issued July 30, 1991 (Ex. 1010,
`“Wrench”).
`
`F. Prior Art and Asserted Grounds
`Petitioner asserts that claims 1–8 of the ’140 patent are unpatentable
`on the following grounds:
`Claim(s) Challenged 35 U.S.C. §
`1–3, 5, 7, 8
`103(a)
`1–3, 5, 7, 8
`103(a)
`4
`103(a)
`4
`103(a)
`6
`103(a)
`
`Reference(s)/Basis
`Jiang, Chen
`Jiang, Chen, Lucke
`Jiang, Chen, Robinson
`Jiang, Chen, Lucke, Robinson
`Jiang, Chen, Wrench
`
`5
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`Claim(s) Challenged 35 U.S.C. §
`6
`103(a)
`
`Reference(s)/Basis
`Jiang, Chen, Lucke, Wrench
`
`Pet. 6. Petitioner relies on the Declaration testimony of Christopher
`Schmandt. Ex. 1003.
`
`II. ANALYSIS
`A. Legal Standards
`“In an [inter partes review], the petitioner has the burden from the
`onset to show with particularity why the patent it challenges is
`unpatentable.” Harmonic Inc. v. Avid Tech., Inc., 815 F.3d 1356, 1363
`(Fed. Cir. 2016) (citing 35 U.S.C. § 312(a)(3) (requiring inter partes review
`petitions to identify “with particularity . . . the evidence that supports the
`grounds for the challenge to each claim”)); see also 37 C.F.R. § 42.104(b)
`(requiring a petition for inter partes review to identify how the challenged
`claim is to be construed and where each element of the claim is found in the
`prior art patents or printed publications relied upon).
`A claim is unpatentable under 35 U.S.C. § 103(a) if “the differences
`between the subject matter sought to be patented and the prior art are such
`that the subject matter as a whole would have been obvious at the time the
`invention was made to a person having ordinary skill in the art to which said
`subject matter pertains.” KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 406
`(2007). The question of obviousness is resolved on the basis of underlying
`factual determinations, including: (1) the scope and content of the prior art;
`(2) any differences between the claimed subject matter and the prior art;
`(3) the level of skill in the art; and (4) when in evidence, objective evidence
`of obviousness or nonobviousness, i.e., secondary considerations. See
`Graham v. John Deere Co., 383 U.S. 1, 17–18 (1966). An obviousness
`
`6
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`analysis “need not seek out precise teachings directed to the specific subject
`matter of the challenged claim, for a court can take account of the inferences
`and creative steps that a person of ordinary skill in the art would employ.”
`KSR, 550 U.S. at 418.
`Additionally, the obviousness inquiry typically requires an analysis of
`“whether there was an apparent reason to combine the known elements in
`the fashion claimed by the patent at issue.” KSR, 550 U.S. at 418 (citing
`In re Kahn, 441 F.3d 977, 988 (Fed. Cir. 2016) (requiring “articulated
`reasoning with some rational underpinning to support the legal conclusion of
`obviousness”)). Furthermore, Petitioner does not satisfy its burden of
`proving obviousness by employing “mere conclusory statements,” but “must
`instead articulate specific reasoning, based on evidence of record, to support
`the legal conclusion of obviousness.” In re Magnum Oil Tools Int’l, Ltd.,
`829 F.3d 1364, 1380 (Fed. Cir. 2016).
`B. Level of Ordinary Skill in the Art
`Petitioner contends that a person having ordinary skill in the art
`(“POSITA”) 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,” and that “[a]dditional education or experience might substitute
`for the above requirements.” Pet. 5 (citing Ex. 1003 ¶¶ 24–26). At this
`stage of the proceeding, “Patent Owner does not dispute the Petition’s
`proposed level of skill for a person of skill in the art.” Prelim. Resp. 4.
`Based on the record presented, including our review of the ’140 patent
`and the types of problems and solutions described in the ’140 patent and
`cited prior art, we agree with Petitioner’s proposed definition of the level of
`
`7
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`ordinary skill in the art and apply it for purposes of this Decision. 1 See, e.g.,
`Ex. 1001,1:21–2:45 (describing various methods of speech recognition
`systems).
`
`C. Claim Interpretation
`We interpret the challenged claims
`using the same claim construction standard that would be used to
`construe the claim in a civil action under 35 U.S.C. 282(b),
`including construing the claim in accordance with the ordinary
`and customary meaning of such claim as understood by one of
`ordinary skill in the art and the prosecution history pertaining to
`the patent.
`37 C.F.R. § 42.100(b) (2022). “In determining the meaning of [a] disputed
`claim limitation, we look principally to the intrinsic evidence of record,
`examining the claim language itself, the written description, and the
`prosecution history, if in evidence.” DePuy Spine, Inc. v. Medtronic
`Sofamor Danek, Inc., 469 F.3d 1005, 1014 (Fed. Cir. 2006). Claim terms
`are given their plain and ordinary meaning as would be understood by a
`person of ordinary skill in the art at the time of the invention and in the
`context of the entire patent disclosure. Phillips v. AWH Corp., 415 F.3d
`1303, 1313 (Fed. Cir. 2005) (en banc). “There are only two exceptions to
`this general rule: 1) when a patentee sets out a definition and acts as his own
`lexicographer, or 2) when the patentee disavows the full scope of a claim
`term either in the specification or during prosecution.” Thorner v. Sony
`Comput. Entm’t Am. LLC, 669 F.3d 1362, 1365 (Fed. Cir. 2012).
`
`
`1 To the extent the parties dispute the level of ordinary skill in the art for the
`’140 patent or the effective filing date of any of the challenged claims, the
`parties are encouraged to address the issue in their papers during trial.
`
`8
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`Petitioner “applies the plain and ordinary meaning of all claim terms
`as understood by a POSITA for all terms.” Pet. 7. Patent Owner contends
`its “arguments in this Preliminary Response are based on the plain and
`ordinary meaning of the words of the claims themselves in view of the
`associated teachings of the specification.” Prelim. Resp. 3.
`We conclude that no terms require express construction at this time.
`See Nidec Motor Corp. v. Zhongshan Broad Ocean Motor Co., 868 F.3d
`1013, 1017 (Fed. Cir. 2017) (“Because we need only construe terms ‘that are
`in controversy, and only to the extent necessary to resolve the controversy,’
`we need not construe [a particular claim limitation] where the construction is
`not ‘material to the . . . dispute.’” (citation omitted)).
`D. Discretionary Denial under 35 U.S.C. § 314(a)
`Institution of inter partes review is discretionary. See Harmonic Inc.
`v. Avid Tech., Inc., 815 F.3d 1356, 1367 (Fed. Cir. 2016) (“[T]he [U.S.
`Patent and Trademark Office (‘USPTO’)] is permitted, but never compelled,
`to institute an IPR proceeding.”). Patent Owner urges us to exercise
`discretion to deny institution of inter partes review under § 314(a) based on
`the factors established in Apple Inc. v. Fintiv Inc., IPR2020-00019, Paper 11
`(PTAB Mar. 20, 2020) (precedential) (“Fintiv”). Prelim. Resp. 5–8.
`Petitioner urges us not to exercise such discretion. Pet. 76–77.
`On June 21, 2022, the Director of the USPTO issued an Interim
`Procedure for Discretionary Denials in AIA Post Grant Proceedings with
`Parallel District Court Litigation, available at
`https://www.uspto.gov/sites/default/files/documents/interim_proc_discretion
`ary_denials_aia_parallel_district_court_litigation_memo_20220621_.pdf
`(the “Interim Procedure”). The Interim Procedure explains that “[c]onsistent
`with Sotera Wireless, Inc., the PTAB will not discretionarily deny institution
`
`9
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`in view of parallel district court litigation where a petitioner presents a
`stipulation not to pursue in a parallel proceeding the same grounds or any
`grounds that could have reasonably been raised before the PTAB.” Interim
`Procedure 3 (internal footnote omitted) (citing Sotera Wireless, Inc. v.
`Masimo Corp., IPR2020-01019, Paper 12 (PTAB Dec. 1, 2020)
`(precedential as to § II.A)). The Interim Procedure explains that such a
`Sotera stipulation “mitigates concerns of potentially conflicting decisions
`and duplicative efforts between the district court and the PTAB . . . . [and]
`allows the PTAB to review grounds that the parallel district court litigation
`will not resolve.” Id. at 7–8.
`In this proceeding, Petitioner stipulates that “[i]f the instant IPR is
`instituted, Petitioner will not pursue in the parallel district court proceeding
`the same grounds as in the Petition or any grounds that could have been
`reasonably raised in the pending Petition.” Paper 9. In light of Petitioner’s
`stipulation, we decline to deny institution under Fintiv. Interim Procedure 3.
`E. Challenge to Independent Claim 1
`Petitioner contends that claim 1 would have been obvious over the
`combined teachings of Jiang and Chen. Pet. 9–44.
`1. Overview of Jiang
`Jiang, entitled “System for Using Silence in Speech Recognition,” is
`related to “computer speech recognition performed by conducting a prefix
`tree search of a silence bracketed lexicon.” Ex. 1004, code (54), 1:16–18.
`
`
`
`10
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`Figure 2 shows a block diagram of speech recognition system 60, and
`is reproduced below.
`
`
`
`Figure 2 above shows microphone 62, analog-to-digital (A/D)
`converter 64, training module 65, feature extraction module 66, silence
`detection module 68, lexicon storage module 70, phonetic speech unit
`storage module 72, tree search engine 74, and output device 76. Id. at 6:20–
`25.
`
`Jiang discloses that the speech recognition system:
`recognizes speech based on an input data stream indicative of the
`speech. Possible words represented by the input data stream are
`provided as a prefix tree including a plurality of phoneme
`branches connected at nodes. The plurality of phoneme branches
`are bracketed by at least one input silence branch corresponding
`to a silence phone on an input side of the prefix tree and at least
`one output silence branch corresponding to a silence phone on an
`output side of the prefix tree.
`Id. at 4:7–16. Lexicon storage module 70 contains information which is
`representative of all of the words in the vocabulary of speech recognition
`system 60. Id. at 7:65–67. The words are presented to tree search engine 74
`in the form of a prefix tree which can be traversed from a root to a leaf to
`
`11
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`arrive at the word most likely indicative of the utterance of the user. Id. at
`7:67–8:4. As the tree is traversed from an input node to an output node, a
`score is assigned to each node connected to a phoneme branch then under
`consideration. Id. at 8:16–20.
`
`A pruning technique can be used by comparing the score at a given
`node with the largest score from the other nodes of a frame being
`considered. Id. at 8:52–55. If the score at that node is sufficiently lower
`than the largest score, that branch is pruned from the tree, thereby drastically
`reducing the search space. Id. at 8:55–64; 11:6–12. A threshold level can be
`implemented for the comparison. Id. at 11:6–8. The lower the threshold
`level, the more branches will be retained throughout the search, and thus the
`more accurate the recognition system will be. Id. at 11:18–21. The
`threshold level can be determined empirically so as to gain an increase in
`computational savings, while significantly reducing the error rate associated
`with the pruning technique. Id. at 11:21–24.
`Overview of Chen
`2.
`Chen “relates generally to parallel processing computer systems for
`performing multiple-instruction-multiple-data (MIMD) parallel processing.”
`Ex. 1005, 1:29–31. Chen describes an architecture for high performance
`MIMD multiprocessors, which organizes the multiprocessors into four or
`more physically separable clusters, and provides for a shared memory model
`to be used with programs executed in the floating shared memory space, and
`a distributed memory model to be used with any programs executed across
`non-adjacently interconnected clusters. Id. at 1:32–43.
`Figure 4 is a block diagram of the architecture of the Chen system,
`annotated by Petitioner, and is reproduced below.
`
`12
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`
`
`Figure 4 above shows computer clusters 100a (in red), 100b (in green),
`100c(in blue), and 100d (in green), processors (P) 102a–d, and memories
`104a–d. Chen’s architecture provides all three of the major memory types
`for parallel processing:
`A true shared memory model with symmetrical access to a
`common shared memory 104a is provided for all processors 102a
`in cluster 100a. An extended shared memory model is provided
`for all of the processors 102a, 102b and 102c that adjacently
`access the cluster shared memories 104a, 104b and 104c in the
`floating shared memory 110, for example. Finally, a distributed
`shared memory model is provided for all processors 102a that
`
`13
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`need to access the cluster shared memory 104d of a non-
`adjacently connected cluster 100d, for example.
`Id. at 9:44–56.
`
`3. Claim 1
`A speech recognition circuit comprising
`The preamble of claim 1 recites a “speech recognition circuit.”
`Petitioner contends that to the extent the preamble is limiting, Jiang
`describes the preamble in disclosing a speech recognition system
`implemented on a computer. Pet. 12–13.
`Based on the evidence and arguments currently of record, for
`purposes of institution, we determine that Petitioner has demonstrated that
`Jiang teaches the features recited in the preamble of claim 1. Because we
`determine that Jiang teaches the preamble, we need not decide whether the
`preamble is limiting at this stage.
`One or more clusters of processors . . . .
`Claim 1 recites “one or more clusters of processors, each of the one or
`more clusters of processors comprising: a plurality of processors.”
`Petitioner contends that the combination of Jiang and Chen teaches this
`limitation. Petitioner contends that Jiang teaches calibrating a pruning
`threshold to ensure that computation time keeps pace with the rate at which
`new frames are received. Pet. 18. Petitioner contends that limited
`computational resources require a high pruning threshold in order to reduce
`the search space. Id. Petitioner contends that the high pruning threshold
`comes at the cost of an increased error rate. Id. Petitioner contends that
`Chen teaches a parallel processing computer system comprising a cluster of
`processors. Id. at 15–17. Petitioner contends that a person of ordinary skill
`would have implemented the speech recognition techniques of Jiang on the
`
`14
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`clustered processors of Chen to efficiently process state sequences of hidden
`Markov models (HMMs) in parallel, in order to improve computational
`resources, allowing the pruning threshold to be relaxed, yielding a decreased
`error rate. Id. at 17–19.
`Patent Owner contends that Chen’s processing architecture is
`incompatible as a substitute for Jiang’s processor. Prelim. Resp. 8. Patent
`Owner contends that Jiang implements its speech recognition techniques on
`a central processing unit (CPU) of a conventional personal computer. Id. at
`9–10. Patent Owner contends that Chen’s system of parallel processing is
`directed towards improving upon prior supercomputers. Id. at 10–11.
`Patent Owner further contends that Chen’s system requires at least eight
`processors and four cluster shared memories, as well as complex
`connections between the processors, resulting in greater cost, complexity,
`and space requirements that what would have been suitable for a
`conventional personal computer at the time. Id. at 11–12.
`Mr. Schmandt testifies that the strictness to which a pruning threshold
`is set in a speech recognition system depends upon the computing resources
`available, where a higher pruning threshold uses less computational
`resources but has a higher chance of errors, and a more relaxed pruning
`threshold has a lower chance of errors but uses more computational
`resources and is thus more expensive. Ex. 1003 ¶ 74. Mr. Schmandt
`testifies that Chen contemplates reducing the cost of a parallel processing
`computing platform as a goal of its invention. Id. Mr. Schmandt testifies
`that modifying Jiang’s speech recognition circuit to use clusters of
`processors as taught by Chen yields the predictable result of providing a
`more powerful computing platform that enables a more relaxed pruning
`
`15
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`threshold, while reducing the expense incurred by grouping commercially
`available, simple processors into clusters. Id.
`On the preliminary evidence, we do not agree with Patent Owner that
`Chen’s system of parallel processing is directed to supercomputers. Chen
`teaches that commercially available single-chip microprocessors and
`memory chips can be used to build its parallel processing computer system.
`Ex. 1005, 10:17–21. We also do not agree with Patent Owner that a person
`of ordinary skill in the art would have regarded the cost of using multiple
`processors to perform Jiang’s techniques of speech recognition to outweigh
`the known benefits. The U.S. Court of Appeals for the Federal Circuit has
`stated that “a given course of action often has simultaneous advantages and
`disadvantages, and this does not necessarily obviate motivation to combine.”
`Medichem, S.A. v. Rolabo, S.L., 437 F.3d 1157, 1165 (Fed. Cir. 2006).
`“Instead, the benefits, both lost and gained, should be weighed against one
`another.” Id. (quoting Winner Int’l Royalty Corp. v. Wang, 202 F.3d 1340,
`1349 n.8 (Fed. Cir. 2000)). At this stage, we agree with Mr. Schmandt, that
`using Chen’s clusters of processors to perform Jiang’s method of speech
`recognition provides the speed and processing power needed to relax the
`pruning threshold while reducing the expense of grouping commercially
`available processors into clusters.
`Patent Owner contends that modifying a standard personal computer
`to implement Chen’s processor and memory architecture would have
`required a level of expertise in parallel computer design that would have
`exceeded the qualifications of a person of ordinary skill. Prelim. Resp. 12.
`Patent Owner contends “an ordinary artisan would not have been motivated
`to substitute Jiang’s generic ‘processing unit 21’/‘CPU 21’ with the Cray
`supercomputer” disclosed in Chen, because such a combination would have
`
`16
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`been the work of a highly unusual artisan using highly unordinary creativity.
`Id. at 13–14. According to Patent Owner, Jiang’s speech recognition
`technique is directed to a CPU which operates in serial mode, and is not
`suited for a parallel processing environment. Id. at 14.
`Mr. Schmandt testifies that many speech recognition systems
`recognized the advantages of employing clusters of processors, such as
`power, flexibility, scalability, and cost management. Ex. 1003 ¶¶ 59–60
`(citing Ex. 1006; Ex. 1024; Ex. 1025; Ex. 1026; Ex. 1027; Ex. 1028). Mr.
`Schmandt testifies that a person of ordinary skill in the art would have been
`familiar with systems using clustered processors for speech recognition and
`would have been familiar with the advantages gained by using a clustered
`processor architecture. Id. at ¶ 60. Mr. Schmandt testifies that a person of
`ordinary skill would have been able to substitute the clustered processing
`architecture of Chen for Jiang’s processing architecture to yield the
`predictable result of improving processing speed and power. Id. at ¶ 67.
`Mr. Schmandt further testifies that a person of ordinary skill would have
`been able to add necessary hardware, such as a microphone, to Chen’s
`computing platform, as well as programming Chen’s platform with speech
`recognition software as taught by Jiang, to yield the predictable results of
`increasing speed and efficiency. Id. at ¶ 68. On this record, we are
`sufficiently persuaded that programming Chen’s computing platform with
`speech recognition software as taught by Jiang was within the level of
`ordinary skill.
`Based on the evidence and arguments currently of record, for
`purposes of institution, we determine that Petitioner has demonstrated that
`the combination of Jiang and Chen teaches the features recited in this
`limitation of claim 1.
`
`17
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`An acoustic model memory storing acoustic model data states
`Claim 1 recites “an acoustic model memory storing acoustic model
`data.” Petitioner contends that Jiang teaches acoustic model data in
`disclosing phonetic speech models such as Hidden Markov Models (HMMs)
`stored in memory. Pet. 19–22. Petitioner contends that Chen teaches that
`each cluster of two or more processors includes a shared memory. Pet. 22–
`24. Petitioner contends that a person of ordinary skill would have stored at
`least a portion of the acoustic model data of Jiang in the shared cluster
`memories of Chen in order to leverage shared memory amongst clustered
`processors during speech recognition. Pet. 24–25.
`Based on the evidence and arguments currently of record, for
`purposes of institution, we determine that Petitioner has demonstrated that
`the combination of Jiang and Chen teaches the features of this limitation of
`claim 1.
`Wherein each of the plurality of processors . . . .
`Claim 1 recites “wherein each of the plurality of processors is
`configured to compute a probability using the acoustic model data in the
`acoustic model memory.” Petitioner contends that Jiang teaches computing
`a probability using the acoustic model data in disclosing determining a most
`likely phoneme based on the stored HMMs. Pet. 25–28. Petitioner contends
`that the combination of Jiang and Chen teaches that each of the plurality of
`processors is configured to compute using the acoustic model data in the
`acoustic model memory. Id. at 28. Petitioner contends that Jiang modified
`by Chen teach an acoustic model, accessible to its plurality of processors,
`used to perform a tree search to determine a most likely phoneme
`represented by features extracted from a user’s utterance. Pet. 31.
`
`18
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`Patent Owner contends the Petition “provides no reason why the
`ordinary artisan would have been motivated to specifically utilize each of
`Chen’s shared cluster memories 104a-d as an ‘acoustic model memory’
`storing ‘acoustic model data,’ as recited in limitations 1(a)-(c) of the
`challenged claims,” other than impermissible hindsight. Prelim. Resp. 18
`(citing Ex. 2002 ¶ 37). Patent Owner contends “an ordinary artisan would
`know that even using each of Chen’s clustered processors for performing the
`speech recognition taught in Jiang would not necessarily lead to each of the
`processors being ‘configured to compute a probability,’ as limitation 1(c)
`requires.” Prelim. Resp. 21.
`Mr. Schmandt testifies that a person of ordinary skill would have
`stored at least a portion of the acoustic model data of Jiang in the shared
`cluster memories of each of the clusters of Chen for access by the
`corresponding processors during speech recognition, to yield the benefits of
`a relaxed pruning threshold at a reduced financial cost. Ex. 1003 ¶ 76. Mr.
`Schmandt testifies that a person of ordinary skill would have recognized that
`it would have been a waste for any particular processor not to be configured
`to compute a probability using the acoustic model data in the acoustic model
`memory. Id. at ¶ 83. Mr. Schmandt testifies that in order to maximize both
`computational and cost efficiency, a person of ordinary skill “would have
`configured the Jiang+Chen modified speech recognition circuit to compute a
`probability using the acoustic model data stored in the acoustic model
`memory.” Id.
`Based on the evidence and arguments currently of record, for
`purposes of institution, we determine that Petitioner has demonstrated that
`the combination of Jiang and Chen teaches the features of this limitation of
`claim 1.
`
`19
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`Wherein: the speech recognition circuit is configured . . . .
`Claim 1 recites “wherein: the speech recognition circuit is configured
`to generate an initial score for an audio sample.” Petitioner contends that
`Jiang describes this limitation in disclosing a tree search engine using
`sampled audio to generate scores as a lexical tree is traversed. Pet. 32–36.
`Based on the evidence and arguments currently of record, for
`purposes of institution, we determine that Petitioner has demonstrated that
`the combination of Smyth and Mozer teaches the features recited in this
`limitation of claim 1.
`And the initial score is used . . . .
`Claim 1 recites “the initial score is used to determine a final score via
`processing a larger amount of model data than was processed to generate the
`initial score.” Petitioner contends that Jiang describes this limitation in
`disclosing using an initial score to determine whether to continue processing
`to determine a final score, where the final score is determined by processing
`more nodes and branches than were processed to generate the initial score.
`Pet. 36–44.
`Based on the evidence and arguments currently of record, for
`purposes of institution, we determine that Petitioner has demonstrated that
`the combination of Jiang and Chen teaches the features of this limitation of
`claim 1. We determine that Petitioner demonstrates a reasonable likelihood
`that independent claim 1 is unpatentable under 35 U.S.C. § 103(a) over Jiang
`and Chen.
`
`F. Additional Grounds and Claims
`Petitioner challenges claims 1–8 as unpatentable over various
`combinations of Jiang, Chen, Lucke, Robinson, and Wrench. Pet. 53–62.
`
`20
`
`
`
`IPR2023-00037
`Patent 10,971,140 B2
`Patent Owner does not separately argue against Petitioner’s showing with
`respect to the additional grounds and claims. See Prelim. Resp.
`At this stage, because Petitioner demonstrates a reasonable likelihood
`of prevailing with respect to its challenge of the independent claim based on
`the combination of Jiang and Chen, we do not see the need to further address
`the challenges of the additional grounds and claims. Further analysis is best
`left for trial after full development of the record. Because “[e]qual treatment
`of claims and grounds for institution purposes has pervasive support in
`SAS,” we institute on all claims and all grounds as challenged in the Petition.
`PGS Geophysical AS v. Iancu, 891 F.3d 13