• Apple Inc. / Amazon Web Services, Inc. et al. v. Zentian LimitedNo claims unpatentable
  PTAB · Jun 11, 2024 · IPR2023-00037 · APJ Jeffrey S Smith · IPR · Final written decision

  Claims 1–8 were not shown unpatentable because Petitioner did not establish that the asserted combinations taught the claimed limitations or provide a persuasive reason to combine Jiang and Chen.

  RulingsFinal written decision

  • BurdenNot shown unpatentable

    “algorithm on Chen’s plurality of clusters of processors, the Petition contends that Jiang alone teaches limitations 1(d) and 1(e) in disclosing assigning an initial score to each of the initial nodes of a prefix tree, then using these initial scores to determine, via pruning, whether to continue processing branches connected to the nodes. Pet. 38 (citing Ex. 1004, 8–64); see id. at 31–44. We agree with Patent Owner that the Petition does not explain how Jiang’s pruning would be implemented in Chen’s multiple clusters of processors to generate an initial score for an audio sample, and to use the initial score “to determine whether to continue processing” as required by limitations 1(d) and 1(e). In contending that Jiang alone teaches limitations 1(d) and 1(e), the Petition quotes from Jiang’s description of its pruning technique:” slip op. at 20

  • Strike new argumentNew argument disregardedDecided for patent owner

    “in the Petition, and is therefore untimely. PO Sur-Reply 4. We agree with Patent Owner. The Petition relies on Jiang alone to teach limitations 1(d) and 1(e), as discussed above. Petitioner’s contention that Jiang’s algorithm would have been replicated across Chen’s processors for performing limitations 1(d) and 1(e) is not found in the Petition. Rather, as discussed above, the Petition describes how Jiang’s algorithm would have been replicated for performing limitations 1(a), 1(b), and 1(c). Petitioner’s new theory in Reply is untimely. A “[p]etitioner may not submit new evidence or argument in reply that it could have presented earlier, e.g., to make out a prima facie case of unpatentability.” Patent Trial and Appeal Board Consolidated Trial Practice Guide (Nov. 2019) 73;2 see also Intelligent Bio-” slip op. at 23

  • Obviousness (§103) › Motivation to combine / reasonable expectation of successMotivation not foundNot shown unpatentable

    “inconsistent with Jiang’s teaching that pruning is performed by comparing a score at a particular node to the maximum score for other corresponding nodes in the prefix tree. Ex. 1004, 8:55–64. We rely on Dr. Anderson’s testimony, that a parallel implementation of Jiang’s algorithm “requires extensive communication between computational components” and that “all four memories 104(a)-(d) would contain necessary information” to perform limitations 1(d) and 1(e), in finding that performing Jiang’s algorithm using Chen’s clusters of processors would have required the processors to access data in all of the cluster memories 104a-d in order to determine the scores of the corresponding nodes of the tree. Ex. 2020 ¶¶ 29, 51.” slip op. at 32

  • Skill levelSkill level petitioner adoptedOther

    “We do not need to resolve this dispute, because even under Petitioner’s definition of the level of ordinary skill, Petitioner has not shown that the challenged claims are unpatentable. We apply Petitioner’s definition that a person having ordinary skill in the art would have had a master’s degree in computer engineering, computer science, electrical engineering, or a related field, with two years of experience in the field of speech recognition, or a bachelor’s degree in the same fields with four years of experience in the field of speech recognition, and that additional education or experience might substitute for the above requirements.” slip op. at 9

  Patent outcomesUS 10,971,140 not shown unpatentable
  § 103 obviousnesscites KSR · Graham
  Patents in suitUS 10,971,140
• Apple Inc. / Amazon Web Services, Inc. et al. v. Zentian LimitedSome claims unpatentable
  PTAB · Jun 11, 2024 · IPR2023-01195 · APJ Jeffrey S Smith · IPR · Final written decision

  Claims 10, 12–14, 16–18, 20–23, 25–28, 35, 37–39, and 42–45 are unpatentable; the remaining challenged claims were not shown unpatentable under the asserted obviousness grounds.

  RulingsFinal written decision

  • Skill levelSkill level board ownDecided for patent owner

    “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. 3 (citing Ex. 1003 ¶¶ 24–25). Patent Owner contends that the Petition and supporting evidence, including the testimony of Mr. Schmandt, does not demonstrate that a person of ordinary skill in the field of speech recognition would have been able to fabricate an integrated circuit. PO Resp. 1. Patent Owner contends that a person of ordinary skill in the field of speech recognition would not also be expected to have specialized knowledge in integrated circuit fabrication in Petitioner does not present arguments to the contrary. See generally Reply. We agree with Petitioner, that a person of ordinary skill in the field of speech recognition would have had a master’s degree in computer engineering, computer science, electrical engineering, or a related field, with two years of experience in the field of speech recognition, or a bachelor’s degree in the same fields with four years of experience in the field of speech recognition.3 Ex. 1001, 1:25–2:27 (describing in the “Background of Invention” section of the ’789 patent various speech recognition systems and methods). We also agree with Patent Owner, and rely on the supporting testimony of Dr. Anderson, in finding that a person of ordinary skill in the field of speech recognition would not have been a person of ordinary skill in the field of integrated circuit fabrication, and would not have had the education, experience, or skill necessary to fabricate an integrated circuit.” slip op. at 8

  • Expert testimonyPetitioner expert discountedPatent owner expert creditedMixed

    “We agree with Patent Owner that Petitioner has not shown, by a preponderance of evidence, that state memory 342 of Smyth is fabricated on the same integrated circuit as processor 341. Petitioner, in asserting that processor 341 and state memory 342 are on the same integrated circuit, cites the following testimony of Mr. Schmandt: In my opinion, the Motorola DSP56000 provides both memory and a processor fabricated on a single integrated circuit (IC). Kloker (Ex. 1009), 1 (discussing the Motorola DSP56000 includes “a core processor, RAM, ROM, peripheral interfaces, and a memory expansion interface on a single, 88-pin chip[”)]. . . . A POSITA would have understood or at least found obvious that Smyth thus teaches a single IC on which both classifier processor 341 and state memory 342 are fabricated. Ex. 1003 ¶ 165; Pet. 39. We do not credit Mr. Schmandt’s testimony because it is inconsistent with Smyth’s disclosure. Smyth discloses that classifier 34 comprises classifying processor 341 and state memory 342, and that “classifying processor 341 may be a suitably programmed digital signal processing (DSP) device.” Ex. 1005, 5:55–57, 6:5–6. Thus, Smyth itself, in discussing classifier 34, distinguishes processor 341 from state memory 342 in stating that only processor 341, and not both processor 341 and state memory 342, may be located on the DSP. describe processor 341 and state memory 342 as being fabricated on a single integrated circuit. Ex. 2020 ¶¶ 34–37. Dr. Anderson persuasively explains that processor 341 and state memory 342 could not be fabricated on the same integrated circuit, because the design of the DSP for processor 341 does not include any memory that meets the requirements of state memory 342. Id. ¶¶ 38–45. We rely on the testimony of Dr. Anderson in finding that Petitioner has not shown that Smyth teaches that processor 341 and state memory 342 are fabricated on a single integrated circuit.” slip op. at 17

  • Strike new argumentNew argument disregardedDecided for patent owner

    “Therefore, we disagree with Petitioner’s contention that “the Petition did not map a POSITA physically fabricating an IC, but rather mapped a single IC on which a calculating apparatus and acoustic model memory were already fabricated.” Reply 23–24. We find that this contention was untimely presented in the Reply. We do not permit Petitioner to propose a mapping of the prior art in the Petition and then completely change that mapping after Patent Owner argues the initial mapping was flawed. A “petitioner may not submit new evidence or argument in reply that it could have presented earlier, e.g., to make out a prima facie case of unpatentability.” TPG 73; see also Intelligent Bio-Sys., Inc. v. Illumina Cambridge Ltd., 821 F.3d 1359, 1369–70 (Fed. Cir. 2016) (holding that the Board did not err in refusing the reply brief as improper under 37 C.F.R. § 42.23(b) because petitioner relied on an entirely new rationale to explain why one of skill in the art would have combined the references at issue); 37 C.F.R. § 42.23(b); Apple Inc. v. eWatch, Inc., IPR2015-00412, Paper 50, 44 (PTAB May 6, 2016) (“‘Respond,’ in the context of 37 C.F.R. § 42.23(b), does not mean embark in a new direction with a new approach as compared to the position originally taken in the Petition. Accepting such belatedly presented new arguments would be unjust to the Patent Owner and we decline to do so.”).” slip op. at 28

  • Obviousness (§103) › Graham factors (scope of art, differences, level of skill)Not shown unpatentableUnpatentableMixed

    “We agree with Patent Owner. Petitioner has not shown that any of the integrated circuits cited in the Petition would have been able to include the claimed acoustic model memory. Further, we find, contrary to Petitioner’s contention, that Patent Owner did not admit that procuring a processor and a memory capable of storing an acoustic model memory fabricated on an integrated circuit was within the level of ordinary skill. To the contrary, throughout the entire proceeding, Patent Owner has consistently argued that the claimed “calculating apparatus and the acoustic model memory are fabricated on a single integrated circuit” was unknown in the prior art. Petitioner has not shown that using the prior art elements relied on in the Petition according to their established functions would have resulted in “the calculating apparatus and the acoustic model memory are fabricated on a single integrated circuit” as recited in claim 1. In summary, Petitioner relies on Mozer to teach the benefits of fabricating a calculating apparatus and an acoustic model memory on the same integrated circuit, but Mozer does not teach a calculating apparatus and an acoustic model memory fabricated on the same integrated circuit. Further, the benefits of Mozer identified by Petitioner are benefits that both Mozer and Smyth already have, without fabricating the calculating apparatus and the acoustic model memory on the same integrated circuit. Also, even if Petitioner had shown there was a reason to combine Smyth and Mozer, Petitioner does not persuasively show that the combination teaches “the calculating apparatus and the acoustic model memory are fabricated on a single integrated circuit” as recited in claim 1. The Petition identifies only model memory for storing an acoustic model defining a plurality of acoustic states” as claimed, which is the one that “a POSITA would have been motivated . . . to fabricate” (Pet. 44), which was beyond the level of ordinary skill. We find that the Petition has not shown that claim 1 is unpatentable over Smyth and Mozer.” slip op. at 38

  Patent outcomesUS 10,839,789 mixed
  § 103 obviousnesscites KSR · Graham
  Patents in suitUS 10,839,789
• Apple Inc. / Amazon Web Services, Inc. et al. v. Zentian LimitedSome claims unpatentable
  PTAB · Jun 11, 2024 · IPR2023-00036 · APJ Jeffrey S Smith · IPR · Final written decision

  Claims 10, 12–14, 16–18, 20–23, 25–28, 35, 37–39, and 42–45 are unpatentable; claims 1, 2, 4, 6–9, 11, 24, and 29 were not shown unpatentable, principally because the asserted combinations did not establish the claimed …

  RulingsFinal written decision

  • Skill levelSkill level petitioner adoptedDecided for patent owner

    “(“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. 3 (citing Ex. 1003 ¶¶ 24–25). Patent Owner contends that the Petition and supporting evidence, including the testimony of Mr. Schmandt, does not demonstrate that a person of ordinary skill in the field of speech recognition would have been able to fabricate an integrated circuit. PO Resp. 1. Patent Owner contends that a person of ordinary skill in the field of speech recognition would not also be expected to have specialized knowledge in integrated circuit fabrication in Petitioner does not present arguments to the contrary. See generally Reply. We agree with Petitioner, that a person of ordinary skill in the field of speech recognition would have had a master’s degree in computer engineering, computer science, electrical engineering, or a related field, with two years of experience in the field of speech recognition, or a bachelor’s degree in the same fields with four years of experience in the field of speech recognition.3 Ex. 1001, 1:25–2:27 (describing in the “Background of Invention” section of the ’789 patent various speech recognition systems and methods). We also agree with Patent Owner, and rely on the supporting testimony of Dr. Anderson, in finding that a person of ordinary skill in the field of speech recognition would not have been a person of ordinary skill in the field of integrated circuit fabrication, and would not have had the education, experience, or skill necessary to fabricate an integrated circuit.” slip op. at 8

  • Claim Construction › Threshold & procedure › Whether a term needs construction (O2 Micro)No construction neededOther

    “Petitioner “applies the plain and ordinary meaning of all claim terms as understood by a POSITA for all terms.” Pet. 5. We conclude that no terms require express construction to resolve a dispute. 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.’” (internal citation omitted)).” slip op. at 10

  • Expert testimonyPetitioner expert discountedPatent owner expert creditedMixed

    “We agree with Patent Owner that Petitioner has not shown, by a preponderance of evidence, that state memory 342 of Smyth is fabricated on the same integrated circuit as processor 341. Petitioner, in asserting that processor 341 and state memory 342 are on the same integrated circuit, cites the following testimony of Mr. Schmandt: In my opinion, the Motorola DSP56000 provides both memory and a processor fabricated on a single integrated circuit (IC). Kloker (Ex. 1009), 1 (discussing the Motorola DSP56000 includes “a core processor, RAM, ROM, peripheral interfaces, and a memory expansion interface on a single, 88-pin chip[”)]. . . . A POSITA would have understood or at least found obvious that Smyth thus teaches a single IC on which both classifier processor 341 and state memory 342 are fabricated. Ex. 1003 ¶ 165; Pet. 39. We do not credit Mr. Schmandt’s testimony because it is inconsistent with Smyth’s disclosure. Smyth discloses that classifier 34 comprises classifying processor 341 and state memory 342, and that “classifying processor 341 may be a suitably programmed digital signal processing (DSP) device.” Ex. 1005, 5:55–57, 6:5–6. Thus, Smyth itself, in discussing classifier 34, distinguishes processor 341 from state memory 342 in stating that only processor 341, and not both processor 341 and state memory 342, may be located on the DSP. describe processor 341 and state memory 342 as being fabricated on a single integrated circuit. Ex. 2020 ¶¶ 34–37. Dr. Anderson persuasively explains that processor 341 and state memory 342 could not be fabricated on the same integrated circuit, because the design of the DSP for processor 341 does not include any memory that meets the requirements of state memory 342. Id. ¶¶ 38–45. We rely on the testimony of Dr. Anderson in finding that Petitioner has not shown that Smyth teaches that processor 341 and state memory 342 are fabricated on a single integrated circuit.” slip op. at 17

  • Strike new argumentNew argument disregardedDecided for patent owner

    “In our Decision to Institute, we understood Petitioner’s theory based on the combination of Smyth and Mozer to include fabricating an integrated circuit in stating that we were sufficiently persuaded “that fabricating a single integrated circuit that includes a calculating apparatus and an acoustic model memory . . . was within the level of ordinary skill.” Dec. 23. Our understanding was based on the Petition’s multiple contentions that a person of ordinary skill would have found it obvious to fabricate Smyth’s processor and state memory on a single integrated circuit. See Pet. 41–44. For example, the Petition contends that “[a] POSITA would have found it obvious . . . to fabricate [Smyth’s processor and state memory] on a single integrated circuit” (Pet. 41–42 (emphasis added)); “a POSITA would have understood and found obvious . . . to fabricate a calculating apparatus and acoustic model memory on a single integrated circuit” (Pet. 42 (emphasis added)); “[s]uch a use of a known technique of fabricating a memory and processor on a single integrated circuit” (id. (emphasis added)); “it would have been obvious to modify Smyth to form the classifier processor and associated memory on a single IC” (Pet. 43 (emphasis added)); (“it was well known in the prior art to fabricate a single integrated circuit” (Pet. 43–44 (emphasis added)); “[t]hus, a POSITA would have been motivated . . . to fabricate [Smyth’s processor and state memory] on a single integrated circuit” (Pet. 44 (emphasis added)). integrated circuit to using integrated circuits that were already fabricated. Reply 22–25. According to Petitioner, “the Petition did not map a POSITA physically fabricating an IC, but rather mapped a POSITA modifying Smyth according to Mozer’s undisputed teachings of a single IC on which a calculating apparatus and acoustic model memory were already fabricated.” Id. at 23–24. However, as discussed in the previous paragraph, when discussing the combination of a processor and an acoustic model memory on an integrated circuit, the Petition used the verb “to fabricate” four times in the future tense and once in the present tense, as well as the verb “to form” once in the future tense. The Petition did not use the past tense phrase “were already fabricated” as alleged in the Reply. Moreover, the Petition did not map Smyth’s state memory (which the Petition contends teaches the claimed “acoustic model memory” (see Pet. 11)) to a memory already fabricated on an integrated circuit in the prior art. Rather, the Petition relied on Mozer to support its contention that a person of ordinary skill would have fabricated the calculating apparatus and acoustic model memory of Smyth on an integrated circuit. See Pet. 42, 44. The Petition cited Boike and Ozcelik in contending that fabricating an integrated circuit with processors and memories was well known, and cited Toyoda in contending that fabricating an integrated circuit with a calculating apparatus and an acoustic model memory was well known. Pet. 43–44 (citing Ex. 1006; Ex. 1029; Ex. 1034). However, the Petition did not explain how a memory located in one of the integrated circuits of Mozer, Boike, Ozcelik, or Toyoda would have been able to store Smyth’s state memory. Therefore, we disagree with Petitioner’s contention that “the Petition did not map a POSITA physically fabricating an IC, but rather mapped a single IC on which a calculating apparatus and acoustic model memory were already fabricated.” Reply 23–24. We find that this contention was untimely presented in the Reply. We do not permit Petitioner to propose a mapping of the prior art in the Petition and then completely change that mapping after Patent Owner argues the initial mapping was flawed. A “petitioner may not submit new evidence or argument in reply that it could have presented earlier, e.g., to make out a prima facie case of unpatentability.” TPG 73; see also Intelligent Bio-Sys., Inc. v. Illumina Cambridge Ltd., 821 F.3d 1359, 1369–70 (Fed. Cir. 2016) (holding that the Board did not err in refusing the reply brief as improper under 37 C.F.R. § 42.23(b) because petitioner relied on an entirely new rationale to explain why one of skill in the art would have combined the references at issue); 37 C.F.R. § 42.23(b); Apple Inc. v. eWatch, Inc., IPR2015-00412, Paper 50, 44 (PTAB May 6, 2016) (“‘Respond,’ in the context of 37 C.F.R. § 42.23(b), does not mean embark in a new direction with a new approach as compared to the position originally taken in the Petition. Accepting such belatedly presented new arguments would be unjust to the Patent Owner and we decline to do so.”).” slip op. at 27

  Patent outcomesUS 10,839,789 mixed
  Claim construction§ 103 obviousnesscites KSR · Graham
  Patents in suitUS 10,839,789
• Apple Inc. / Amazon Web Services, Inc. et al. v. Zentian LimitedAll claims unpatentable
  PTAB · Jun 11, 2024 · IPR2023-01193 · APJ Christopher L Ogden · IPR · Final written decision

  Claims 1, 4, 5, 7, 9, 10, 12, and 14–16 are unpatentable as obvious under § 103(a) over the listed Jiang-based combinations.

  RulingsFinal written decision

  • Skill levelSkill level petitioner adoptedDecided for petitioner

    “Relying on Mr. Schmandt’s testimony, Apple argues that a person of ordinary skill in the art would have had “a master’s degree in computer engineering, computer science, electrical engineering, or a related field, with . . . two years of experience in the field of speech recognition, or a bachelor’s degree in the same fields with . . . four years of experience in the field of speech recognition.” Pet. 3 (citing Ex. 1003 ¶¶ 28–30). Zentian does not contest Apple’s proposed definition of the level of ordinary skill. PO Resp. 5. Because Apple’s articulation of that level of skill is uncontested, supported by testimonial evidence, and appears consistent with the types of problems and solutions in the ’277 patent, we adopt it for this decision. See, e.g., Ex. 1001, 1:4–8:10.” slip op. at 7

  • Claim Construction › Threshold & procedure › Whether a term needs construction (O2 Micro)No construction neededNot reached

    “Apple argues that all terms should be given their plain and ordinary meaning, except that the limitation “calculating means for calculating a distance indicating the similarity between a feature vector and a predetermined acoustic state of an acoustic model” in claim 14 (Ex. 1001, 39:6–8) is a means-plus-function limitation and thus invokes 35 U.S.C. § 112 para. 6 (2006). Pet. 5–6. Zentian does not contest these arguments or propose any explicit constructions. See PO Resp. We do not find it necessary to explicitly construe any terms for this decision. To the extent that we need to interpret claim language, we address that language below in the context of the prior art. See Nidec Motor Corp. v. Zhongshan Broad Ocean Motor Co., 868 F.3d 1013, 1017 (Fed. Cir. 2017) (“[W]e need only construe terms ‘that are in controversy, and only to the extent necessary to resolve the controversy’ . . . .” (quoting Vivid Techs., Inc. v. Am. Sci & Eng’g, Inc., 200 F.3d 795, 803 (Fed. Cir. 1999))).” slip op. at 9

  • Expert testimonyPetitioner expert creditedPatent owner expert discountedDecided for petitioner

    “We find Apple’s arguments persuasive that a person of ordinary skill in the art would have understood Jiang’s pruning technique not to rely on feedback, particularly because Jiang makes no mention of feedback, and Jiang as a whole appears consistent with Apple’s theory that Jiang does not teach feedback-based pruning. See, e.g., Ex. 1004, 2:5–7, 8:52–64, 10:65– 11:15. We credit Mr. Schmandt’s testimony and supporting evidence that “[e]ven though Jiang still describes calculating all scores initially (for each state for example), Jiang explains that pruning thereafter saves on computational resources.” Ex. 1003 ¶ 194 (citing Ex. 1004, 10:65–11:25). We find this testimony more persuasive than Mr. Anderson’s contrary testimony which does not credibly find support in the text of Jiang. See Ex. 2020 ¶¶ 57–71. We also disagree that Apple’s arguments that Jiang’s distance calculations do not rely on pruning are untimely. PO Resp. 19. The Petition argues that Jiang teaches calculating distances, “namely a score for each phoneme branch of the prefix tree.” Pet. 21 (citing Ex. 1004, 8:29–33). Also, Mr. Schmandt clearly opined, in his initial declaration submitted with the Petition, that “each phoneme branch is scored” in Jiang’s search tree. Ex. 1003 ¶ 149. Thus, Zentian had clear notice in the Petition of Apple’s argument that Jiang performs distance calculations for the entire search tree. Apple’s later argument that this process does not necessitate feedback is a natural consequence of its argument in the Petition, and is not untimely.” slip op. at 27

  • Obviousness (§103) › Motivation to combine / reasonable expectation of successUnpatentableMotivation foundUnpatentable

    “(1) Apple’s Proposed Combination According to Apple, Brown teaches that processing stages in a speech-recognition system may be pipelined such that “as the frames of data are generated, the data is processed from one stage to another so that the operations are synchronized with the rate of frame generation.” Pet. 44–45 (citing Ex. 1036, 3:32–48; Ex. 1003 ¶¶ 206–209). Apple contends that a person of ordinary skill in the art would have had reason to modify the combination of Jiang and Baumgartner by pipelining data sent from the CPU to the SLA and then back to the CPU because “[p]ipelining data offered numerous well-known advantages including greater throughput of the processor, improved performance, decreased cost, and simplified programming.” Pet. 45 (citing Ex. 1003 ¶ 210). Apple also argues that a person of ordinary skill in the art “would have realized pipelining keeps all processors operating efficiently in a multi-processor system, such that wasteful idle-processing time is minimized.” Pet. 46 (citing Ex. 1003 ¶¶ 212–213). And according to Apple, “Jiang already discusses the importance of time with respect to pruning.” Pet. 46 (citing Ex. 1004, 10:65– 11:25). Apple also argues that a person of ordinary skill in the art would have had a reasonable expectation of success in adding Brown’s pipelining to the Jiang–Baumgartner combination because “[p]ipelining was well-known in the art including in speech recognition systems,” and the three references “teach speech recognition systems using similar hardware, such as various processors to perform similar speech recognition functions,” such that the combination “would have been a trivial modification.” Pet. 46–47 (citing Ex. 1004, 4:60–65, 6:32–36, 6:39–42; Ex. 1007 ¶ 33, Fig. 19; Ex. 1036, 3:11– 15; Ex. 1003 ¶ 214).” slip op. at 21

  Patent outcomesUS 7,979,277 unpatentable
  Claim construction§ 103 obviousnesscites Phillips · KSR · Graham
  Patents in suitUS 7,979,277
• Apple Inc. / Amazon Web Services, Inc. et al. v. Zentian LimitedAll claims unpatentable
  PTAB · Jun 11, 2024 · IPR2023-00034 · APJ Christopher L Ogden · IPR · Final written decision

  Claims 1, 4, 5, 7, 9, 10, 12, and 14–16 are unpatentable as obvious under § 103(a) over the identified combinations.

  RulingsFinal written decision

  • Skill levelSkill level petitioner adoptedDecided for petitioner

    “Relying on Mr. Schmandt’s testimony, Apple argues that a person of ordinary skill in the art would have had “a master’s degree in computer engineering, computer science, electrical engineering, or a related field, with . . . two years of experience in the field of speech recognition, or a bachelor’s degree in the same fields with . . . four years of experience in the field of speech recognition.” Pet. 3 (citing Ex. 1003 ¶¶ 28–30). Zentian does not contest Apple’s proposed definition of the level of ordinary skill. PO Resp. 5. Because Apple’s articulation of that level of skill is uncontested, supported by testimonial evidence, and appears consistent with the types of problems and solutions in the ’277 patent, we adopt it for this decision. See, e.g., Ex. 1001, 1:4–8:10.” slip op. at 7

  • Claim Construction › Threshold & procedure › Whether a term needs construction (O2 Micro)No construction neededDecided for patent owner

    “Apple argues that all terms should be given their plain and ordinary meaning, except that the limitation “calculating means for calculating a distance indicating the similarity between a feature vector and a predetermined acoustic state of an acoustic model” in claim 14 (Ex. 1001, 39:6–8) is a means-plus-function limitation and thus invokes 35 U.S.C. § 112 para. 6 (2006). Pet. 5–6. Zentian does not contest these arguments or propose any explicit constructions. See PO Resp. We do not find it necessary to explicitly construe any terms for this decision. To the extent that we need to interpret claim language, we address that language below in the context of the prior art. See Nidec Motor Corp. v. Zhongshan Broad Ocean Motor Co., 868 F.3d 1013, 1017 (Fed. Cir. 2017) (“[W]e need only construe terms ‘that are in controversy, and only to the extent necessary to resolve the controversy’ . . . .” (quoting Vivid Techs., Inc. v. Am. Sci & Eng’g, Inc., 200 F.3d 795, 803 (Fed. Cir. 1999))).” slip op. at 9

  • Obviousness (§103) › Motivation to combine / reasonable expectation of successMotivation foundReasonable expectation foundDecided for petitioner

    “We find Apple’s arguments persuasive that pruning is not required as part of its proposed combination of Jiang, Baumgartner, and Brown. Nothing in Jiang suggests that pruning is mandatory—it is presented as option, possibly a preferred option, but not a requirement. See Ex. 1004, 8:52 (“[A] pruning technique can be used.” (emphasis added)). Although Apple relied on the possibility of pruning in Jiang as part of the motivation to combine Jiang’s system with Baumgartner’s parallel-processing hardware, Apple never included pruning as an element needed to meet any limitation of claim 1. See Pet. 41, 46; Ex. 1003 ¶¶ 194, 201, 213. Although Jiang teaches the desirability of using pruning, at least “for a large vocabulary” (Ex. 1004, 2:19), there is nothing in claim 1 limiting how small the acoustic model can be. Also, whether or not a person of ordinary skill in the art could have achieved real-time speech-recognition system without pruning is not pertinent to achieving the claimed invention because claim 1 does not require real-time speech recognition. Apple has persuasively shown that a person of ordinary skill in the art would have had reason to implement Jiang’s system on parallel-processing hardware, as Baumgartner teaches, using pipelining as Brown teaches to meet all the limitations of claim 1, even without Jiang’s pruning embodiment, and that this combination would have been within the ordinary skill in the art.” slip op. at 23

  • Expert testimonyPetitioner expert creditedPatent owner expert discountedDecided for petitioner

    “We find Apple’s arguments persuasive that a person of ordinary skill in the art would have understood Jiang’s pruning technique not to rely on feedback, particularly because Jiang makes no mention of feedback, and Jiang as a whole appears consistent with Apple’s theory that Jiang does not teach feedback-based pruning. See, e.g., Ex. 1004, 2:5–7, 8:52–64, 10:65– 11:15. We credit Mr. Schmandt’s testimony and supporting evidence that “[e]ven though Jiang still describes calculating all scores initially (for each state for example), Jiang explains that pruning thereafter saves on computational resources.” Ex. 1003 ¶ 194 (citing Ex. 1004, 10:65–11:25). We find this testimony more persuasive than Mr. Anderson’s contrary testimony which does not credibly find support in the text of Jiang. See Ex. 2020 ¶¶ 57–71. We also disagree that Apple’s arguments that Jiang’s distance calculations do not rely on pruning are untimely. PO Resp. 19. The Petition argues that Jiang teaches calculating distances, “namely a score for each phoneme branch of the prefix tree.” Pet. 21 (citing Ex. 1004, 8:29–33). Also, Mr. Schmandt clearly opined, in his initial declaration submitted with the Petition, that “each phoneme branch is scored” in Jiang’s search tree. Ex. 1003 ¶ 149. Thus, Zentian had clear notice in the Petition of Apple’s argument that Jiang performs distance calculations for the entire search tree. Apple’s later argument that this process does not necessitate feedback is a natural consequence of its argument in the Petition, and is not untimely.” slip op. at 27

  Patent outcomesUS 7,979,277 unpatentable
  Claim construction§ 103 obviousnesscites Phillips · KSR · Graham
  Patents in suitUS 7,979,277
• Apple Inc. / Amazon Web Services, Inc. et al. v. Zentian LimitedAll claims unpatentable
  PTAB · Jun 11, 2024 · IPR2023-01192 · APJ Christopher L Ogden · IPR · Final written decision

  The Board held claims 46–51, 53, 54, 58, 59, 64, and 67 unpatentable as obvious under § 103 over the specified combinations of Thelen, Bailey, Chen, and additional references.

  RulingsFinal written decision

  • Expert testimonyPetitioner expert creditedPatent owner expert discountedDecided for petitioner

    “We find Apple’s arguments persuasive notwithstanding Zentian’s arguments to the contrary. We credit Mr. Schmandt’s testimony that consolidating the overlapping data within similar lexical models (such as “sports” models like football or baseball) and storing the consolidated data in a shared memory that multiple processors may access in parallel would have resulted in memory and processing efficiency. Ex. 1003 ¶¶ 83–84. This would have been an improvement over Thelen, where each processor has access to only one, distinct lexical model at a time and there is no shared memory in which multiple processors can access the same consolidated models data in parallel. See Ex. 1030, 7:30–34.11” slip op. at 27

  • Skill levelSkill level petitioner adoptedDecided for petitioner

    “Because a person of ordinary skill in the art is a hypothetical person, we agree with Apple that Dr. Anderson’s inability to implement a particular computer vision application within the constraints of a particular Texas Instruments parallel-processing environment does not mean that more general knowledge and skill in parallel processing would have been beyond the level of ordinary skill. To the contrary, the evidence of record suggests that a person of ordinary skill in the art at the time of the claimed invention, who, for example, had a master’s degree and two years of experience in the field of speech recognition, would have also had knowledge and experience with parallel processing.” slip op. at 9

  • Claim Construction › Threshold & procedure › Whether a term needs construction (O2 Micro)No construction neededOther

    “Apple states that it “applies the plain and ordinary meaning of all claim terms,” and does not propose any explicit constructions. See Pet. 7. Likewise, Zentian does not propose any explicit constructions. See PO Resp. We do not find it necessary to explicitly construe any terms for this decision. To the extent that we need to interpret claim language, we address that language below in the context of the prior art. See Nidec Motor Corp. v. Zhongshan Broad Ocean Motor Co., 868 F.3d 1013, 1017 (Fed. Cir. 2017) (“[W]e need only construe terms ‘that are in controversy, and only to the extent necessary to resolve the controversy’ . . . .” (quoting Vivid Techs., Inc. v. Am. Sci & Eng’g, Inc., 200 F.3d 795, 803 (Fed. Cir. 1999))).” slip op. at 11

  • Obviousness (§103) › Motivation to combine / reasonable expectation of successMotivation foundUnpatentable

    “would not have been dissuaded from the memory-efficiency benefits of modifying Thelen–Bailey by grouping processors and memory into clusters because memory collisions would have allegedly made recognition slower. Although memory collisions may have been an issue for the ordinarily skilled artisan to consider, Apple has persuasively shown that there are other motivating considerations, such as the known flexibility12 and scalability of distributed processing systems in general, decreased cost because off-the-shelf components may be used, and the added efficiency of data sharing among related sub-context models. See Pet. 40–45. Apple persuasively shows, and we find Mr. Schmandt’s supporting testimony credible, that a skilled artisan would have expected to be able to minimize the effect of memory collisions based on Chen’s teachings and the background knowledge in the art. See Pet. 14–19. For the same reasons, we do not find credible Dr. Anderson’s testimony (see, e.g., Ex. 2020 ¶¶ 44–46)” slip op. at 28

  Patent outcomesUS 7,587,319 unpatentable
  Claim construction§ 103 obviousnesscites Phillips · KSR · Graham
  Patents in suitUS 7,587,319
• Apple Inc. / Amazon Web Services, Inc. et al. v. Zentian LimitedAll claims unpatentable
  PTAB · Jun 11, 2024 · IPR2023-00033 · APJ Christopher L Ogden · IPR · Final written decision

  Claims 46–51, 53, 54, 58, 59, 64, and 67 are unpatentable as obvious under § 103 over the listed combinations of prior art.

  RulingsFinal written decision

  • Skill levelSkill level petitioner adoptedDecided for petitioner

    “Because a person of ordinary skill in the art is a hypothetical person, we agree with Apple that Dr. Anderson’s inability to implement a particular computer vision application within the constraints of a particular Texas Instruments parallel-processing environment does not mean that more general knowledge and skill in parallel processing would have been beyond the level of ordinary skill. To the contrary, the evidence of record suggests that a person of ordinary skill in the art at the time of the claimed invention, who, for example, had a master’s degree and two years of experience in the field of speech recognition, would have also had knowledge and experience with parallel processing.” slip op. at 9

  • Expert testimonyPetitioner expert creditedDecided for petitioner

    “We find Apple’s arguments persuasive notwithstanding Zentian’s arguments to the contrary. We credit Mr. Schmandt’s testimony that consolidating the overlapping data within similar lexical models (such as “sports” models like football or baseball) and storing the consolidated data in a shared memory that multiple processors may access in parallel would have resulted in memory and processing efficiency. Ex. 1003 ¶¶ 83–84. This would have been an improvement over Thelen, where each processor has access to only one, distinct lexical model at a time and there is no shared memory in which multiple processors can access the same consolidated models data in parallel. See Ex. 1030, 7:30–34.11” slip op. at 27

  • Claim Construction › Threshold & procedure › Whether a term needs construction (O2 Micro)No construction neededOther

    “Apple states that it “applies the plain and ordinary meaning of all claim terms,” and does not propose any explicit constructions. See Pet. 7. Likewise, Zentian does not propose any explicit constructions. See PO Resp. We do not find it necessary to explicitly construe any terms for this decision. To the extent that we need to interpret claim language, we address that language below in the context of the prior art. See Nidec Motor Corp. v. Zhongshan Broad Ocean Motor Co., 868 F.3d 1013, 1017 (Fed. Cir. 2017) (“[W]e need only construe terms ‘that are in controversy, and only to the extent necessary to resolve the controversy’ . . . .” (quoting Vivid Techs., Inc. v. Am. Sci & Eng’g, Inc., 200 F.3d 795, 803 (Fed. Cir. 1999))).” slip op. at 11

  • Obviousness (§103) › Motivation to combine / reasonable expectation of successMotivation foundDecided for petitioner

    “We find Apple’s arguments persuasive notwithstanding Zentian’s arguments to the contrary. We credit Mr. Schmandt’s testimony that consolidating the overlapping data within similar lexical models (such as “sports” models like football or baseball) and storing the consolidated data in a shared memory that multiple processors may access in parallel would have resulted in memory and processing efficiency. Ex. 1003 ¶¶ 83–84. This would have been an improvement over Thelen, where each processor has access to only one, distinct lexical model at a time and there is no shared memory in which multiple processors can access the same consolidated models data in parallel. See Ex. 1030, 7:30–34.11 11 The dissent characterizes Thelen’s common storage 340 as a shared memory. However, each of Thelen’s recognizers only has access (indirectly The evidence of record suggests that the ordinarily skilled artisan would not have been dissuaded from the memory-efficiency benefits of modifying Thelen–Bailey by grouping processors and memory into clusters because memory collisions would have allegedly made recognition slower. Although memory collisions may have been an issue for the ordinarily skilled artisan to consider, Apple has persuasively shown that there are other motivating considerations, such as the known flexibility12 and scalability of distributed processing systems in general, decreased cost because off-the-shelf components may be used, and the added efficiency of data sharing among related sub-context models. See Pet. 40–45.” slip op. at 27

  Patent outcomesUS 7,587,319 unpatentable
  Claim construction§ 103 obviousnesscites Phillips · KSR · Graham
  Patents in suitUS 7,587,319
• Apple Inc. / Amazon Web Services, Inc. et al. v. Zentian LimitedAll claims unpatentable
  PTAB · Jun 11, 2024 · IPR2023-01194 · APJ Kevin F Turner · IPR · Final written decision

  Claims 1–6 are unpatentable as obvious: claims 1–5 over their listed combinations of Jiang, Smyth, Nguyen, Boike, or Baumgartner, and claim 6 over Jiang, Smyth, Nguyen, Boike, and Baumgartner.

  RulingsFinal written decision

  • Skill levelSkill level petitioner adoptedDecided for petitioner

    “person of ordinary skill in the art at the time of the invention “would have had a master’s degree in computer engineering, computer science, electrical engineering, or a related field,” and would have had “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.” Pet. 4 (citing Ex. 1003 ¶ 24). Patent Owner indicates that it does not “dispute the Petition’s proposed level of skill for a person of skill in the art,” but also emphasizes that such artisans who possess “ordinary skill in the field of speech recognition, [would] not [be] one of extraordinary creativity.” PO Resp. 4. We continue to determine, as we did in the Institution Decision, that Petitioner’s proposal consistent with the level of ordinary skill in the art reflected by the prior art of record, see Inst. Dec. 7; Okajima v. Bourdeau, 261 F.3d 1350, 1355 (Fed. Cir. 2001); In re GPAC Inc., 57 F.3d 1573, 1579 continue to adopt Petitioner’s unopposed position as to the level of ordinary skill in the art for purposes of this decision.” slip op. at 8

  • Claim Construction › Intrinsic evidence (Phillips canons) › Plain and ordinary meaning to a POSITATerm construed board ownDecided for petitioner

    “of that claim against the prior art.” In re Wilson, 424 F.2d 1382, 1385 (CCPA 1970). The recitation of claim 1 that FVs are extracted and/or derived means that FVs may not only be extracted from the digital audio stream, thus obtaining a FV based on the scale of the FV (“representing a point in an N-dimensional space,” depending on the value of N (Ex. 1001 13:19–23)). Such FVs, per claim 1, may also be representations of the digital audio stream, i.e., quantities derived therefrom. Any such derived quantities would perhaps even be lossy. See Ex. 1001, 13:13–18. As such, FVs, according to independent claim 1, are not limited to pure feature vectors, which are quantities extracted from the digital audio stream, but also include quantities derived from the stream, which are representations” slip op. at 11

  • Obviousness (§103) › Graham factors (scope of art, differences, level of skill)UnpatentableUnpatentable

    “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 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. “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 the adjudicator to provide “some 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).” slip op. at 7

  • Obviousness (§103) › Motivation to combine / reasonable expectation of successMotivation foundDecided for petitioner

    “above, we determine that Jiang uses codewords, which are representative feature vectors within the scope of claim 1, and those codewords are used to calculate distances. The Petition discusses a motivation to modify Jiang to use Smyth’s DSP device to perform distance calculations (Pet. 40-41), but it does so in the context of “[t]o the extent Patent Owner argues that Jiang does not teach calculating distances.” Id. at 40. Given that we determine that Jiang does teach performing distancing calculations that meet the claim limitations, Patent Owner’s contention is not persuasive. Rather, we are persuaded by Petitioner’s motivation for using Smyth’s dedicated DSP to perform Jiang’s distance calculations, namely, for the benefit of allowing for segmentation of higher-load processing, resulting in a faster, more efficient, ¶¶ 176–180; Ex. 1004, 6:39–42). We agree with Mr. Schmandt, that “[i]t would have been obvious to modify Jiang’s tree search engine, which performs both distance calculation and word determination, to include Smyth’s distinct processors for performing each respective functionality,” because such a modification “results in a more efficient, faster, and more accurate speech recognition process by segmenting higher-load processing tasks.” Ex. 1003 ¶ 180 (citing Ex. 1012, 19, Ex. 1005, 5:49–53, 6:5–8, 6:43–51; Ex. 1004, 4:60–5:3). Additionally, Patent Owner argues that using Smyth’s probability calculation in the system of Jiang will dramatically increase the computation burden of the calculations, and this is contrary to the purpose of Jiang to reduce computational requirements, as evinced by the use of lossy codewords. PO Resp. 16–19 (citing Ex. 1002 ¶¶ 54–59). Petitioner asserts, with respect to this latter argument, that “the Board need not consider this argument if the Board determines that Jiang teaches limitation 1(g).” Pet. Reply 14. We agree with Petitioner. Given the discussion with respect to limitation 1(g) above, and that Jiang is sufficient to teach or suggest element 1(g), we need not further consider the combination of Smyth and Jiang in the claimed calculation of distances. Therefore, we determine that the motivation proffered by the Petition is sufficient to combine the teachings of Jiang and Smyth, as disclosed in the Petition.” slip op. at 30

  Patent outcomesUS 10,062,377 unpatentable
  Claim construction§ 103 obviousnesscites KSR · Graham
  Patents in suitUS 10,062,377
• Apple Inc. / Amazon Web Services, Inc. et al. v. Zentian LimitedAll claims unpatentable
  PTAB · Jun 11, 2024 · IPR2023-00035 · APJ Kevin F Turner · IPR · Final written decision

  Claims 1–6 are unpatentable as obvious: claims 1–6 under the listed combinations of Jiang, Smyth, and the additional references for dependent claims.

  RulingsFinal written decision

  • Claim Construction › Intrinsic evidence (Phillips canons) › Plain and ordinary meaning to a POSITATerm construed board ownDecided for petitioner

    “of that claim against the prior art.” In re Wilson, 424 F.2d 1382, 1385 (CCPA 1970). The recitation of claim 1 that FVs are extracted and/or derived means that FVs may not only be extracted from the digital audio stream, thus obtaining a FV based on the scale of the FV (“representing a point in an N-dimensional space,” depending on the value of N (Ex. 1001 13:19–23)). Such FVs, per claim 1, may also be representations of the digital audio stream, i.e., quantities derived therefrom. Any such derived quantities would perhaps even be lossy. See Ex. 1001, 13:13–18. As such, FVs, according to independent claim 1, are not limited to pure feature vectors, which are quantities extracted from the digital audio stream, but also include quantities derived from the stream, which are representations 8). Given such a definition, vector quantization, which reduces the plurality of quantities in a FV to a single codeword that is associated with a codebook entry, representing multiple values, would also comprise “derived” representations of the “quantities from said digital audio stream” as recited in the claim. Although Patent Owner’s declarant, Dr. Anderson, provides that “[c]odewords are the result of vector quantization (VQ) of the original feature vectors, and are thus not the feature vectors themselves” (Ex. 2020 ¶ 39), that comports with one alternative of what encompasses a FV as recited in claim 1, namely, “extracted quantities,” but not the more expansive definition that includes “derived quantities” provided in independent claim 1. The vector quantization of a feature vector results in a vector quantized feature vector, or codeword, that still represents a multitude of values “derived . . . from said digital audio stream,” but is more computationally expedient. See Ex. 1003 ¶¶ 140–141. We agree with Dr. Anderson that persons of ordinary skill in the art would have understood that calculating distances from a feature vector comprising extracted quantities was distinct from computing probabilities from a codeword (Ex. 2020 ¶ 43), but that, again, does not take into consideration the broader definition of “feature vector” confirmed by the claims. Rather, we agree with Petitioner and Mr. Schmandt, that a person of ordinary skill in the art would have understood that a codeword is a representative feature vector. Pet. 22 (citing Ex. 1003 ¶¶ 140–141). A codeword, or representative feature vector, “comprises a plurality of . . . derived,” or vector quantized, “quantities from said digital audio stream” within the meaning of claim 1. claim 1, comprises a plurality of extracted qualities from the digital audio stream during the defined audio time frame, and/or derived quantities that are representative of the digital audio stream during the defined audio time frame, such as the derived values of a codeword.” slip op. at 11

  • Skill levelSkill level petitioner adoptedDecided for petitioner

    “Level of Ordinary Skill in the Art Petitioner, supported by Mr. Schmandt’s testimony, proposes that a person of ordinary skill in the art at the time of the invention “would have had a master’s degree in computer engineering, computer science, electrical engineering, or a related field,” and would have had “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.” Pet. 4 (citing Ex. 1003 ¶ 24). Patent Owner indicates that it does not “dispute the Petition’s proposed level of skill for a person of skill in the art,” but also emphasizes that such artisans who possess “ordinary skill in the field of speech recognition, [would] not [be] one of extraordinary creativity.” PO Resp. 4. We continue to determine, as we did in the Institution Decision, that Petitioner’s proposal consistent with the level of ordinary skill in the art reflected by the prior art of record, see Inst. Dec. 7; Okajima v. Bourdeau, 261 F.3d 1350, 1355 (Fed. Cir. 2001); In re GPAC Inc., 57 F.3d 1573, 1579 continue to adopt Petitioner’s unopposed position as to the level of ordinary skill in the art for purposes of this decision.” slip op. at 8

  • Obviousness (§103) › Motivation to combine / reasonable expectation of successMotivation foundDecided for petitioner

    “We are not persuaded by Patent Owner’s arguments. As discussed above, we determine that Jiang uses codewords, which are representative feature vectors within the scope of claim 1, and those codewords are used to calculate distances. The Petition discusses a motivation to modify Jiang to use Smyth’s DSP device to perform distance calculations (Pet. 40-41), but it does so in the context of “[t]o the extent Patent Owner argues that Jiang does not teach calculating distances.” Id. at 40. Given that we determine that Jiang does teach performing distancing calculations that meet the claim limitations, Patent Owner’s contention is not persuasive. Rather, we are persuaded by Petitioner’s motivation for using Smyth’s dedicated DSP to perform Jiang’s distance calculations, namely, for the benefit of allowing for segmentation of higher-load processing, resulting in a faster, more efficient, ¶¶ 176–180; Ex. 1004, 6:39–42). We agree with Mr. Schmandt, that “[i]t would have been obvious to modify Jiang’s tree search engine, which performs both distance calculation and word determination, to include Smyth’s distinct processors for performing each respective functionality,” because such a modification “results in a more efficient, faster, and more accurate speech recognition process by segmenting higher-load processing tasks.” Ex. 1003 ¶ 180 (citing Ex. 1012, 19, Ex. 1005, 5:49–53, 6:5–8, 6:43–51; Ex. 1004, 4:60–5:3). Additionally, Patent Owner argues that using Smyth’s probability calculation in the system of Jiang will dramatically increase the computation burden of the calculations, and this is contrary to the purpose of Jiang to reduce computational requirements, as evinced by the use of lossy codewords. PO Resp. 16–19 (citing Ex. 1002 ¶¶ 54–59). Petitioner asserts, with respect to this latter argument, that “the Board need not consider this argument if the Board determines that Jiang teaches limitation 1(g).” Pet. Reply 14. We agree with Petitioner. Given the discussion with respect to limitation 1(g) above, and that Jiang is sufficient to teach or suggest element 1(g), we need not further consider the combination of Smyth and Jiang in the claimed calculation of distances. Therefore, we determine that the motivation proffered by the Petition is sufficient to combine the teachings of Jiang and Smyth, as disclosed in the Petition.” slip op. at 30

  • Expert testimonyPetitioner expert creditedPatent owner expert discountedDecided for petitioner

    “Weighing the evidence proffered by both parties, we are not persuaded that Jiang discloses feedback-based pruning. Petitioner’s declarant disputes that Jiang teaches feedback-based pruning (Ex. 2016, 18:16–19:14), and Jiang itself, within its entire discussion of pruning (Ex. 1004, 8:52–64, 10:65–11:15), never discusses feedback of pruned results. We are left with Patent Owner’s assertions that such feedback-based pruning is implicitly taught by Jiang, and we determine that to be insufficient to require a disclosure that is not explicit. We acknowledge that Patent Owner’s declarant (Ex. 2020 ¶ 64–76) provides testimony, based on the teachings of Jiang and contemporaneous teachings in the art, that Jiang’s pruning requires feedback from the search stage to the calculation stage, but we are not persuaded that Jiang must operate with such feedback. We are also persuaded by Petitioner’s point that Jiang computes a score for each phoneme branch considered in the tree, that contradicts Patent Owner’s theory that scores are computed for only active, non-pruned phoneme Lastly, Patent Owner seems to rely, at least partially, that “Jiang never states it does not use feedback-based pruning” (PO Sur-reply 25), but an absence of evidence is not evidence of absence. As such, given that we are not persuaded that Jiang must employ feedback-based pruning, we are not persuaded that the methods of Nguyen cannot be used in the combination of Jiang and Smyth, as proffered by the Petition with respect to claims 2 and 3. In addition, we agree with Petitioner that Jiang discloses that “a pruning technique can be used,” and find that Jiang’s pruning technique is optional. Pet. Reply 16; Ex. 1004, 8:52–64. Although Patent Owner contends that Mr. Schmandt includes Jiang’s pruning as part of his combination theory (PO Resp. 20–22), we agree with Petitioner that the Petition does not rely on Jiang’s pruning technique in mapping Jiang’s process to the search stage of independent claim 1. Pet. Reply 16; see Pet. 44–61. We find that the Petition does not rely on the portions of Mr. Schmandt’s testimony that consider Jiang’s pruning. Further, even under Mr. Schmandt’s combination theory, Mr. Schmandt testifies that the pruning “threshold value can be set at any desired level, based primarily on desired memory and computational savings, and a desired error rate increase caused by memory and computational savings.” Ex. 1003 ¶ 174 (quoting Ex. 1004, 2:25–46). Mr. Schmandt testifies that “if the speech recognition system is programmed to carry more hypotheses forward in the tree (because a relatively lower threshold value is set), the system has a higher recognition accuracy.” Id. ¶ 177. Mr. Schmandt testifies that modifying Jiang’s tree search engine 74 to comprise two digital signal processors, where a first processor performs distance calculations and the second performs other speech recognition tasks, yields the benefit of given that using two digital signal processors allows the first processor to increase accuracy by reducing the pruning threshold, and given that the pruning threshold can be set to any desired level based on a desired error rate, a person of ordinary skill in the art would have set the pruning threshold to zero for the benefit of increasing recognition accuracy and lowering the error rate as much as possible. With the pruning threshold set to zero, the combination does not perform pruning, even under Mr. Schmandt’s combination theory.” slip op. at 35

  Patent outcomesUS 10,062,377 unpatentable
  Claim construction§ 103 obviousnesscites KSR · Graham
  Patents in suitUS 10,062,377
• Intel Corporation / Dell Technologies Inc. et al. v. 3d Surfaces, LLCAll claims unpatentable
  PTAB · Apr 16, 2024 · IPR2023-00004 · APJ George R Hoskins · IPR · Final written decision

  Claims 1–4, 9–11, 17, and 18 are unpatentable: claims 1–4, 9, 11, and 17 as obvious over Sfarti and Foley; claims 10 and 18 as obvious over Sfarti, Foley, and Möller.

  RulingsFinal written decision

  • Skill levelSkill level petitioner adoptedDecided for petitioner

    “to the ’299 patent “would have had at least a Bachelor’s degree in computer science, computer engineering, electrical engineering, or a related scientific field, and approximately two years of work experience in computer graphics or, alternatively, two years[’] work towards a masters or doctorate degree in these scientific fields.” Pet. 10; Ex. 1003 ¶¶ 96–97. Patent Owner applies Petitioner’s formulation of ordinary skill in the Patent Owner Response. See PO Resp. 10–11; Ex. 2005 ¶ 31. Based on the record presented, we now find as we did in the Institution Decision that Petitioner’s formulation of the level of ordinary skill in the art is consistent with the level of skill reflected in the ’299 patent and the prior art of record, and is not disputed by Patent Owner. See Inst. Dec. 20–21. We, therefore, apply Petitioner’s formulation in this Decision.” slip op. at 9

  • Claim Construction › Functional claiming / §112(f) means-plus-function › Whether §112(f) applies (Williamson, nonce words)§112(f) appliesTerm construed board ownDecided for petitioner

    “function limitations subject to § 112, ¶ 6, we conclude that they are. Because the Unit Terms do not use the word “means,” we apply a rebuttable presumption that § 112, ¶ 6 does not apply. See Williamson v. Citrix Online, LLC, 792 F.3d 1339, 1348–49 (Fed. Cir. 2015) (en banc in relevant part). Nonetheless, as Petitioner argues, the claim term “unit” does not connote any particular structure, and the terms “transform unit”, “lighting unit”, “renderer unit”, and “tessellate unit” recite function without reciting sufficient structure for performing that function. The “unit” term at issue here is closely akin to the “module” term at issue in Williamson, and similarly generic terms such as “element” and “device,” and is, therefore, in the context of the ’299 patent claims and specification, a “nonce” word used “in a manner that is tantamount to using the word ‘means.’” Williamson, 792 F.3d at 1350–51; see also Diebold Nixdorf, Inc. v. Int’l Trade Comm’n, 899 F.3d 1291, 1298–1301 (Fed. Cir. 2018) (holding the claim term “cheque standby unit” was subject to § 112, ¶ 6, because it failed to “recite any structure, much less ‘sufficiently definite structure,’” and the claim recited a function without reciting sufficient structure for performing that function, despite an expert witness’s contrary testimony); WSOU Invs. LLC v. Google LLC, No. 2022-1066, 2023 WL 6210607, at *4 (Fed. Cir. Sept. 25, 2023) (stating “the term ‘unit,’ in a vacuum, is defined only by the function it performs, much like terms such as ‘element’ or ‘means’” and “[a] ‘unit’ could be almost anything”).4 We therefore agree with Petitioner’s position that the rebuttable presumption against application of § 112, ¶ 6 is overcome in this case, because the Unit Terms are not understood by persons of ordinary skill in the art to have a sufficiently definite meaning as the name for structure. See Williamson, 792 F.3d at 1349 (“The standard is whether the words of the claim are understood by persons of ordinary skill in the art to have a sufficiently definite meaning as the name for structure.”).” slip op. at 15

  • Claim Construction › Intrinsic evidence (Phillips canons) › Plain and ordinary meaning to a POSITATerm construed board ownDecided for patent owner

    “transforming is performed by a GPU with computer graphics pipeline hardware including a processor programmed to execute the transforming step of the disclosed microcode. See Ex. 1001, 5:20–29, 5:51–67, 7:1–8, 7:12–13, 12:24–28; Ex. 1003 ¶¶ 104–105. We construe “transform unit” as a means-plus-function limitation with the function of transforming graphic objects into transformed objects (claims 1 and 11) including control points (claims 2 and 11), and with the corresponding structure of a GPU with computer graphics pipeline hardware including a processor programmed to execute the transforming step of the microcode disclosed in the specification. The foregoing construction of “transform unit” is identical to the preliminary construction in the Institution Decision, with two exceptions. First, it adds that the corresponding structure is a GPU with computer graphics pipeline hardware including a processor. See Inst. Dec. 22–23. We agree with Patent Owner’s contention that the ’299 patent describes such pipeline hardware as being part of “[t]he present invention,” so we include it in our construction. Ex. 1001, 5:20–29. Second, our construction does not require a dedicated processor, which is part of Petitioner’s claim construction. See Inst. Dec. 22–23; Pet. 11–12. As Patent Owner points out, the ’299 patent specification does not include the term “dedicated.” To the extent Petitioner’s “dedicated” term is meant to require that the processor is programmed solely to perform the claimed transforming function, to the exclusion of any other processing, we conclude the claim is open to both a dedicated processor and a non-dedicated processor.” slip op. at 19

  • Claim Construction › Threshold & procedure › Whether a term needs construction (O2 Micro)No construction neededOther

    “7. Remaining Claim Terms Upon review of the arguments and evidence, we determine no further construction of any claim term is needed. See Nidec Motor Corp. v. Zhongshan Broad Ocean Motor Co., 868 F.3d 1013, 1017 (Fed. Cir. 2017) (per curiam) (claim terms need to be construed “only to the extent necessary to resolve the controversy” (quoting Vivid Techs., Inc. v. Am. Sci. & Eng’g, Inc., 200 F.3d 795, 803 (Fed. Cir. 1999))).” slip op. at 25

  Patent outcomesUS 7,245,299 unpatentable
  Claim construction§ 103 obviousnesscites Graham
  Patents in suitUS 7,245,299

Holdings, summaries, rulings, issues, outcomes, winners and claim constructions shown for a decision were generated by a large language model reading the opinion. They are a guide to the decision, not the decision. Confirm every one against the court's own text before relying on it. Outcomes read from the patent owner’s side. Party names are extracted from the decision and are missing on about 900 short Director denials and scans, which carry the trial number instead.