`U.S. Patent No. 7,979,277
`
`Oral Argument, March 12, 2024
`
`Apple Inc. v. Zentian Limited
`Case No. IPR2023-00034
`
`Petitioner’s Demonstrative Exhibits – Not Evidence
`
`Petitioner’s DX-1
`
`IPR2023-00034
`Apple EX1068 Page 1
`
`
`
`Claim 1 and Proposed Ground 1
`
`- Ground 1: Claim 1
`Jiang (Ex. 1004) in view of Baumgartner (Ex. 1007) and Brown (Ex. 1036)
`-
`
` Claim 1:
`
`Petition (Pet.) (Paper 1), 10, 76; ’277 Patent (Ex. 1001), 38:10-27
`
`Petitioner’s DX-2
`
`
`
`Zentian’s Sole Argument
`
`- Zentian’s sole argument is that feedback-based pruning
`cannot be accomplished with the claimed pipelining
`
`POR, 1
`
`Patent Owner’s Response (POR) (Paper 19), 1
`
`Petitioner’s DX-3
`
`
`
`Pruning at the Search Stage Does Not Require
`Feedback-Based Pruning
`
`- Zentian’s Sur-Reply:
`
`- Schmandt’s Supp. Dec. Opinions:
`
`PO Sur-Reply (Paper 27), 1
`
`Schmandt
`Supp. Dec.,
`¶ 14
`
`Pet. Reply, 2; Schmandt Supp. Dec. (Ex. 1063), ¶ 14
`
`Petitioner’s DX-4
`
`
`
`Zentian’s Argument Requires Bodily
`Incorporation of Feedback in the Combination
`
`Feedback-Based
`Pruning Is:
`
`Not Claimed
`
`Not Mapped in
`the Combination
`
`Not Required to
`Be Performed
`When Pruning
`During the
`Search Stage
`
`Not Taught
`in Jiang
`
`Pet. Reply (Paper 21), 1-5
`
`Petitioner’s DX-5
`
`
`
`Obviousness of Forgoing Feedback-Based
`Pruning for Pipelining
`
`- Mathew & Ravishankar Both Forgo Feedback in Favor
`of Pipelining
`- Krishna Implemented Feedback Due to Its “Generic
`Architecture”
`
`Krishna (Ex. 2018), p. 69
`
`Pet. Reply, 21-27, Pinpoint at 22-25; Schmandt Supp. Dec. (Ex. 1063), ¶¶ 10-11;
`Mathew (Ex. 1050), 4, 11; Ravishankar (Ex. 1021), 245, 109
`
`Petitioner’s DX-6
`
`
`
`Zentian’s Numerous False Assumptions
`
`Zentian Falsely Assumes:
`
`The Mapped Combination Requires Jiang’s Pruning at the Search Stage
`
`Jiang “Implicitly Teaches” Feedback-Based Pruning
`
`The Combination Requires Feedback-Based Pruning Because It Is a Known
`Technique
`
`Pipelining with Feedback-Based Pruning Is Impossible
`
`The Challenged Claims Require Real-Time Speech Recognition
`
`Pet. Reply, 3-5, 5-12, 12-13, 14-19, 20-21
`
`Petitioner’s DX-7
`
`
`
`False Assumption 4:
`Pipelining with Feedback-Based Pruning Is Impossible
`
`- Two Tradeoffs with Pipelining with Feedback-Based
`Pruning
`
`1. Classification performed with one-frame delay
`Result: Classifier may perform unneeded distance calculations if
`-
`phoneme transition
`
`2.
`
`If phoneme transition, there may be a new phoneme
`not on the active list yet
`Result: New phoneme added to active list for next frame with no
`impact on speech recognition
`
`-
`
`Pet. Reply, 16-17; Schmandt Supp. Dec., ¶ 24
`
`Petitioner’s DX-8
`
`
`
`False Assumption 4 – Cont.:
`Pipelining with Feedback-Based Pruning Is Impossible
`
`- No Delay in Speech Recognition, per Mr. Schmandt
`
`Pet. Reply, 16-17
`
`Schmandt Supp. Dec., Ex. 1063, ¶ 21
`
`Petitioner’s DX-9
`
`
`
`False Assumption 4 – Cont.:
`Pipelining with Feedback-Based Pruning Is Impossible
`
`- Brown’s Teachings:
`
`Brown (Ex. 1036), 3:44-47
`
`Pet., 44; Schmandt Declaration (Ex. 1003), ¶ 209
`
`Petitioner’s DX-10
`
`
`
`False Assumption 4 – Cont.:
`Pipelining with Feedback-Based Pruning Is Impossible
`
`Pet. Reply, 16; Schmandt Supp. Dec., ¶ 23
`
`Petitioner’s DX-11
`
`
`
`False Assumption 4 – Cont.:
`Pipelining with Feedback-Based Pruning Is Impossible (Part 1)
`
`Pet. Reply, 16-17; Schmandt Supp. Dec., ¶ 54
`
`Petitioner’s DX-12
`
`
`
`False Assumption 4 – Cont.:
`Pipelining with Feedback-Based Pruning Is Impossible (Part 2)
`
`Pet. Reply, 16-17; Schmandt Supp. Dec., ¶ 25
`
`Petitioner’s DX-13
`
`
`
`False Assumption 1:
`The Combination Requires Jiang’s Pruning at the Search Stage
`
`- Jiang Teaches Pruning Is Optional
`
`- Petition Mapped:
`
`Jiang (Ex. 1004), 8:52
`
`Pet. Reply, 3, 4
`
`Pet. Reply, 4
`
`Petitioner’s DX-14
`
`
`
`False Assumption 1:
`The Combination Requires Jiang’s Pruning at the Search Stage
`
`- Mr. Schmandt’s Deposition Testimony:
`
`Schmandt Depo. Tr. (Ex. 2016), 15:4-9
`
`Pet. Reply, 4; POR, 6; PO Sur-Reply (Paper 27), 12
`
`Petitioner’s DX-15
`
`
`
`False Assumption 2:
`Jiang “Implicitly Teaches” Feedback-Based Pruning Via “Under
`Consideration” Language
`- Jiang Teaches:
`
`Jiang, 8:29-33
`
`Jiang, 8:46-49
`
`Pet. Reply, 7
`
`Jiang, 8:16-20
`
`Jiang, 8:33-37
`
`Petitioner’s DX-16
`
`
`
`False Assumption 2 – Cont.:
`Jiang Feedback-Based Pruning Via “No…Further Processing”
`Disclosure
`Jiang Discloses:
`
`Zentian Argues:
`“pruning eliminates all scoring related to the
`pruned branches and nodes (including at the
`calculation stage).”
`
`Jiang, 8:62-64
`
`POR, 19-20
`
`But Dr. Anderson Admitted:
`“And so when you prune a branch, it will take
`a
`swath of
`them potentially out of
`consideration. If that particular phoneme
`was under consideration in another
`branch,
`it’s
`not
`taken
`out
`of
`consideration.”
`(Ex.
`1064,
`45:16-20
`(emphasis added)).
`
`Pet. Reply, 9-10
`POR, 19-20; Jiang, 8:62-64, Fig. 6; Pet. Reply, 9-10; Anderson Depo. Tr. (Ex. 1064), 45:16-20
`
`Pet. Reply, 10; Jiang, Fig. 6 (Annotated by Petitioner)
`
`Petitioner’s DX-17
`
`