IPR2023-00035
`Apple Inc.
`v.
`Zentian Limited
`Patent 10,062,377
`
`Patent Owner’s Demonstratives
`
`Presented March 12, 2024
`
`1
`
`
`Apple Inc.
`v.
`Zentian Limited
`Patent 10,062,377
`
`Patent Owner’s Demonstratives
`
`Presented March 12, 2024
`
`1
`
`
`
`Argument Roadmap
`
`v Jiang alone cannot meet “feature vector” requirements
`v Petitioner has failed to prove a motivation for modifying
`Jiang to use feature vector-based recognition
`v Petition’s claim 2 theory lacks a reasonable
`expectation of success
`v Petitioner’s new claim 2 theory is improper and refuted
`
`2
`
`
`
`Argument Roadmap
`
`v Jiang alone cannot meet “feature vector” requirements
`v Petitioner has failed to prove a motivation for modifying
`Jiang to use feature vector-based recognition
`v Petition’s claim 2 theory lacks a reasonable
`expectation of success
`v Petitioner’s new claim 2 theory is improper and refuted
`
`3
`
`
`
`Jiang alone cannot meet “feature vector” requirements
`
`A codeword is not a plurality of extracted and/or derived quantities
`
`1(b): “feature vector comprises a
`plurality of extracted and/or derived
`quantities. . .”
`
`v “the n amplitudes for the given
`time interval represent n-
`component vector in the n-
`dimensional space” Nadas, Ex.
`1019, 1:15-37
`
`v Feature vectors had, e.g., thirty-
`nine dimensions. Ex. 1003 ¶¶ 41,
`66, 124
`
`Vector quantized codeword is a single
`quantity
`
`v VQ “code[s] a spectral vector into one of a
`fixed number of discrete symbols” Rabiner,
`Ex. 1015, 28, 35
`
`v VQ converts “multidimensional” input value
`to “single dimension identifying nearest
`vector-quantized value,” i.e., “single value
`associated with the codebook entry”
`Schmandt, Ex. 1010, 60
`
`v VQ converts feature vector to “single lookup
`number or index or code word index or code
`word” Anderson, Ex. 1073, 54:1-3
`
`Sur-reply at 2-3
`
`4
`
`
`
`Jiang alone cannot meet “feature vector” requirements
`
`A codeword is not a plurality of extracted and/or derived quantities
`
`Sur-reply at 2-3; Ex. 1073 at 53:13-54:11
`
`5
`
`
`
`Jiang alone cannot meet “feature vector” requirements
`
`A codeword is not a plurality of extracted and/or derived quantities
`
`Sur-reply at 2-3; POR at 7-8, 16-17; Ex. 1012 at 29 (pdf); Ex. 2005 at 17:34-36
`
`6
`
`
`
`Jiang alone cannot meet “feature vector” requirements
`
`A codeword is not extracted or derived from the recited digital audio stream
`
`1(b): “feature vector comprises . . .
`extracted and/or derived quantities
`from said digital audio stream during
`a defined audio time frame”
`
`v “When an unknown speech input
`is uttered, for each time interval, a
`value is measured or computed for
`each of the n components, where
`each component is referred to as a
`feature. The values of all of the
`features are consolidated to form
`an n-component feature vector for
`a time interval.” Nadas, Ex. 1019,
`1:15-37
`
`Vector quantized codeword is created
`from training data, not audio stream
`
`v VQ codebook (and codewords) developed
`from a “training set” of vectors; Rabiner, Ex.
`1015, 155-156
`
`v Codebook values are created from
`“specimen data” or “sample data”;
`Schmandt, Ex. 1010, 60
`
`v “the codeword is representative of the
`template vector,” which is “based on the
`centroid of other data, data that has come
`and gone and not used in the recognition
`process. It’s, rather, training data.”
`Anderson, Ex. 1073, 54:12-15, 56:4-23
`
`Sur-reply at 4-5
`
`7
`
`
`
`Jiang alone cannot meet “feature vector” requirements
`
`Claims require using distances calculated from feature vectors to identify words
`
`Limitation 1(g): “wherein said identification of spoken
`words uses one or more distances calculated from a first
`feature vector”
`
`Limitation 1(h): “a search stage for using the calculated
`distances to identify words. . .”
`
`Sur-reply at 6-7
`
`8
`
`
`
`Jiang alone cannot meet “feature vector” requirements
`
`Petition only relies on Jiang’s codeword-based recognition to meet 1(g) and 1(h)
`
`Sur-reply at 6-7 (Pet. 56, 59-60)
`
`9
`
`
`
`Jiang alone cannot meet “feature vector” requirements
`
`Jiang only teaches codeword-based word recognition
`
`Sur-reply at 6-7 (Jiang, 8:16-51); Ex. 2020 ¶ 50
`
`10
`
`
`
`Jiang alone cannot meet “feature vector” requirements
`
`“Feature vector” as recited in challenged claims cannot encompass codewords
`
`v Claims define “feature vector”
`
`v Codeword is not a “feature vector”
`
`v “Calculating distances from a
`feature vector” does not include
`computing probabilities from
`codewords. Ex. 2020 ¶ 43; Ex.
`1015, 152-154.
`
`v Calculations from codewords will
`be different than calculations from
`feature vectors. Schmandt, Ex.
`1010, 61-62 (light bulb example);
`Ex. 2020 ¶¶ 43-44
`
`v Known tradeoffs in speech
`recognition, including with respect
`to using VQ, cannot expand claim
`scope beyond express terms
`
`v Specification’s general discussion
`of lossy compression does not
`outweigh claim language
`(limitations 1(b), 1(g), 1(h))
`
`v Specification only teaches
`calculating distances from feature
`vectors, not codewords. Ex. 2020
`¶¶ 45-47, Ex. 1001, 12:42-44,
`13:5-7, 24:62-64, 25:18-20,
`25:57-60, 34:48-50, Fig. 18-23.
`
`Sur-reply at 8-13
`
`11
`
`
`
`Argument Roadmap
`
`v Jiang alone cannot meet “feature vector” requirements
`v Petitioner has failed to prove a motivation for modifying
`Jiang to use feature vector-based recognition
`v Petition’s claim 2 theory lacks a reasonable
`expectation of success
`v Petitioner’s new claim 2 theory is improper and refuted
`
`12
`
`
`
`No motivation to modify Jiang to use
`feature vector-based word recognition
`
`Schmandt dec. ¶¶ 170-182 directed to different
`modification, i.e., adding second processor/DSP to Jiang
`in view of Smyth
`
`Schmandt dec. ¶ 183 provides no motivation for
`modifying Jiang to use feature vector-based word
`recognition
`
`Schmandt dec. ¶ 172 states calculating distances was
`“computationally-intensive,” which motivates against
`modifying Jiang to use more computationally intensive
`feature-vector based word recognition. Ex. 2020 ¶¶ 54-58
`
`Sur-reply at 13-16
`
`13
`
`
`
`No motivation to modify Jiang to use
`feature vector-based word recognition
`
`Petitioner’s assertion that POSA may have determined the
`modification provides “a worthy tradeoff” or that
`increased load “can be offset by a second processor”
`proves no motivation to combine
`
`Sur-reply at 16-17
`
`14
`
`
`
`Argument Roadmap
`
`v Jiang alone cannot meet “feature vector” requirements
`v Petitioner has failed to prove a motivation for modifying
`Jiang to use feature vector-based recognition
`v Petition’s claim 2 theory lacks a reasonable
`expectation of success
`v Petitioner’s new claim 2 theory is improper and refuted
`
`15
`
`
`
`Combination of Jiang and Nguyen lacks
`reasonable expectation of success
`
`Undisputed that Nguyen cannot be combined with feedback-based pruning
`
`v Petitioner’s claim 2 theory relies on
`Nguyen’s approach of “performing
`a similarity measure (i.e. distance
`calculation) while
`contemporaneously performing the
`search step.” Pet. 62; POR, 20.
`
`v Nguyen’s relied upon teaching
`cannot be combined with
`feedback-based pruning. POR 30-
`32, Ex. 2020 ¶¶ 83-88, Ex. 2018
`at 69, Ex. 1050 at 4, Ex. 1012 at
`91.
`
`v Petitioner’s Reply does not mention
`Nguyen and concedes that
`Nguyen’s teaching could not have
`been combined with feedback-
`based pruning
`
`v Petitioner’s “one frame delay”
`theory is not supported by any
`prior art, and is therefore
`irrelevant. Ex. 2021, 55:4-8
`
`Sur-reply at 17-19
`
`16
`
`
`
`Combination of Jiang and Nguyen lacks
`reasonable expectation of success
`
`Jiang’s pruning is part of the combination, and not optional therein
`
`v Ravishankar and Mathew do not
`obviate the need for Jiang’s pruning in
`the combination; Mr. Schmandt
`excluded both from his combination. Ex.
`2021, 41:2-6, 38:11-16
`
`v Patent Owner’s arguments are not
`based on bodily incorporation; they are
`based on the incompatibility between
`Jiang’s pruning teachings and Nguyen’s
`parallel processing teachings
`
`v Petitioner’s combination includes
`Jiang’s pruning technique. POR, 20; Ex.
`2016, 13:19-14:8, Ex. 1003 ¶ 178
`
`v Mr. Schmandt’s supplemental
`declaration did not dispute that the
`combination includes Jiang’s pruning
`
`v Jiang’s pruning is not “optional” in the
`combination because Mr. Schmandt’s
`theory included it
`
`v Mr. Schmandt admitted that reducing
`Jiang’s pruning, much less eliminating
`it, risks the system falling “increasingly
`behind real time.” Ex. 1003 ¶ 177; Ex.
`1004, 2:19-28.
`
`Sur-reply at 19-22
`
`17
`
`
`
`Combination of Jiang and Nguyen lacks
`reasonable expectation of success
`
`Jiang’s pruning is feedback-based pruning
`
`Sur-reply at 19-22
`
`18
`
`
`
`Combination of Jiang and Nguyen lacks
`reasonable expectation of success
`
`Jiang’s pruning is feedback-based pruning
`
`Sur-reply at 19-22; Ex. 1005 at 8:16-20, 8:44-51, 8:52-64, 8:25-43
`
`19
`
`
`
`Combination of Jiang and Nguyen lacks
`reasonable expectation of success
`
`Jiang’s pruning is feedback-based pruning
`
`Sur-reply at 19-22
`
`20
`
`
`
`Combination of Jiang and Nguyen lacks
`reasonable expectation of success
`
`Jiang’s pruning is feedback-based pruning
`
`Krishna, Ex. 2018, at 42
`
`Mathew I, Ex. 1050, at 11
`
`Sur-reply at 19-22; Ex. 1050 at 11; Ex. 2018 at 42
`
`21
`
`
`
`Argument Roadmap
`
`v Jiang alone cannot meet “feature vector” requirements
`v Petitioner has failed to prove a motivation for modifying
`Jiang to use feature vector-based recognition
`v Petition’s claim 2 theory lacks a reasonable
`expectation of success
`v Petitioner’s new claim 2 theory is improper and refuted
`
`22
`
`
`
`Petitioner’s new claim 2 theory is untimely and refuted
`
`“Rather than explaining how its original petition was
`correct, Continental’s subsequent arguments amount to an
`entirely new theory of prima facie obviousness absent from
`the petition. Shifting arguments in this fashion is foreclosed
`by statute, our precedent, and Board guidelines.”
`
`Wasica v. Con’tl Auto Sys., 853 F.3d 1272, 1286-87 (Fed. Cir. 2017)
`
`Sur-reply at 26-28
`
`23
`
`
`
`Petitioner’s new claim 2 theory is untimely and refuted
`
`Sur-reply at 26-28; Ex. 2021, at 43:3-7, 41:2-6, 38:11-16
`
`24
`
`
`
`Petitioner’s new claim 2 theory is untimely and refuted
`
`Sur-reply at 26-28; Ex. 2018 at 42, 69, 108, 142-145
`
`25
`
`
Accessing this document will incur an additional charge of $.
After purchase, you can access this document again without charge.
Accept $ Charge
Still Working On It
This document is taking longer than usual to download. This can happen if we need to contact the court directly to obtain the document and their servers are running slowly.
Give it another minute or two to complete, and then try the refresh button.
A few More Minutes ... Still Working
It can take up to 5 minutes for us to download a document if the court servers are running slowly.
Thank you for your continued patience.
This document could not be displayed.
We could not find this document within its docket. Please go back to the docket page and check the link. If that does not work, go back to the docket and refresh it to pull the newest information.
Your account does not support viewing this document.
You need a Paid Account to view this document. Click here to change your account type.
Your account does not support viewing this document.
Set your membership
status to view this document.
With a Docket Alarm membership, you'll
get a whole lot more, including:
- Up-to-date information for this case.
- Email alerts whenever there is an update.
- Full text search for other cases.
- Get email alerts whenever a new case matches your search.
One Moment Please
The filing “” is large (MB) and is being downloaded.
Please refresh this page in a few minutes to see if the filing has been downloaded. The filing will also be emailed to you when the download completes.
Your document is on its way!
If you do not receive the document in five minutes, contact support at support@docketalarm.com.
Sealed Document
We are unable to display this document, it may be under a court ordered seal.
If you have proper credentials to access the file, you may proceed directly to the court's system using your government issued username and password.
Access Government Site
