·1· · · · ·UNITED STATES PATENT AND TRADEMARK OFFICE
`
`Page 1
`
`·2· · · · · BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`·3
`
`·4· ·APPLE INC.,· · · · · · · )
`· · · · · · · · · · · · · · · )
`·5· · · · · · Petitioner,· · ·)
`· · · · · · · · · · · · · · · ) Case No. IPR2023-00033
`·6· · ·vs.· · · · · · · · · · )
`· · · · · · · · · · · · · · · ) U.S. Patent No. 7,587,319
`·7· ·ZENTIAN LIMITED,· · · · ·)
`· · · · · · · · · · · · · · · )
`·8· · · · · · Patent Owner.· ·)
`
`·9
`
`10
`
`11
`· · VIDEOCONFERENCE DEPOSITION OF DAVID V. ANDERSON, PH.D.
`12
`
`13· · · · · · · TAKEN ON BEHALF OF THE PETITIONER
`
`14
`· · · · · · · · · · · ·NOVEMBER 17, 2023
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22· Court Reporter:
`· · Lauren N. Lawrence, RPR, CCR-MO, CCR-KS
`23· Lexitas Legal
`· · 1608 Locust Street
`24· Kansas City, Missouri 64108
`· · 816.221.1160
`25· 800.280.3376
`
`IPR2023-00037
`Apple EX1035 Page 1
`
`
`
`Page 1
`
`·2· · · · · BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`·3
`
`·4· ·APPLE INC.,· · · · · · · )
`· · · · · · · · · · · · · · · )
`·5· · · · · · Petitioner,· · ·)
`· · · · · · · · · · · · · · · ) Case No. IPR2023-00033
`·6· · ·vs.· · · · · · · · · · )
`· · · · · · · · · · · · · · · ) U.S. Patent No. 7,587,319
`·7· ·ZENTIAN LIMITED,· · · · ·)
`· · · · · · · · · · · · · · · )
`·8· · · · · · Patent Owner.· ·)
`
`·9
`
`10
`
`11
`· · VIDEOCONFERENCE DEPOSITION OF DAVID V. ANDERSON, PH.D.
`12
`
`13· · · · · · · TAKEN ON BEHALF OF THE PETITIONER
`
`14
`· · · · · · · · · · · ·NOVEMBER 17, 2023
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22· Court Reporter:
`· · Lauren N. Lawrence, RPR, CCR-MO, CCR-KS
`23· Lexitas Legal
`· · 1608 Locust Street
`24· Kansas City, Missouri 64108
`· · 816.221.1160
`25· 800.280.3376
`
`IPR2023-00037
`Apple EX1035 Page 1
`
`
`
`·1· · · · · · · · · · · I N D E X
`
`Page 2
`
`·2· · WITNESS:· · · · · · · · · · · · · · · · · · PAGE
`
`·3· · DAVID V. ANDERSON, PH.D.
`
`·4· ·EXAMINATION BY MS. BAILEY· · · · · · · · · · · ·5
`
`·5
`
`·6
`
`·7· · · · · · ·(No exhibits were marked.)
`
`·8
`
`·9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`IPR2023-00037
`Apple EX1035 Page 2
`
`
`
`·1· · · · ·UNITED STATES PATENT AND TRADEMARK OFFICE
`
`Page 3
`
`·2· · · · · BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`·3
`
`·4· ·APPLE INC.,· · · · · · · · )
`· · · · · · · · · · · · · · · · )
`·5· · · · · · Petitioner,· · · ·)
`· · · · · · · · · · · · · · · · )· Case No. IPR2023-00033
`·6· · ·vs.· · · · · · · · · · · )
`· · · · · · · · · · · · · · · · )· U.S. Patent No. 7,587,319
`·7· ·ZENTIAN LIMITED,· · · · · ·)
`· · · · · · · · · · · · · · · · )
`·8· · · · · · Patent Owner.· · ·)
`
`·9
`
`10· · · · · · VIDEOCONFERENCE DEPOSITION OF DAVID V.
`
`11· ANDERSON, PH.D., produced, sworn, and examined on
`
`12· November 17, 2023, all parties appearing remotely,
`
`13· before Lauren N. Lawrence, RPR, CCR-MO, CCR-KS, in a
`
`14· certain cause now pending in the United States Patent
`
`15· and Trademark Office before the Patent Trial and
`
`16· Appeal Board, between APPLE INC., Petitioner, vs.
`
`17· ZENTIAN LIMITED, Patent Owner; on behalf of the
`
`18· Petitioner.
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`IPR2023-00037
`Apple EX1035 Page 3
`
`
`
`·1· · · · · · · · · ·A P P E A R A N C E S
`
`Page 4
`
`·2
`
`·3· APPEARING FOR PETITIONER:
`
`·4· · · ·Ms. Jennifer C. Bailey
`· · · · ·ERISE IP, P.A.
`·5· · · ·7015 College Boulevard
`· · · · ·Suite 700
`·6· · · ·Overland Park, Kansas 66211
`· · · · ·813.777.5600
`·7· · · ·jennifer.bailey@eriseip.com
`
`·8· · · ·Ms. Christina Canino, pro hac vice
`· · · · ·717 17th Street
`·9· · · ·Suite 1400
`· · · · ·Denver, Colorado 80202
`10· · · ·913.777.5600
`
`11
`· · APPEARING FOR PATENT OWNER:
`12
`· · · · ·Mr. Kayvan B. Noroozi
`13· · · ·NOROOZI PC
`· · · · ·11601 Wilshire Boulevard
`14· · · ·Suite 2170
`· · · · ·Los Angeles, California 90025
`15· · · ·310.975.7074
`· · · · ·kayvan@noroozipc.com
`16
`
`17· ALSO PRESENT:
`
`18· · · ·Adam Sandwell, Technical Analyst
`
`19
`
`20
`
`21
`
`22· Court Reporter:
`· · Lauren N. Lawrence, RPR, CCR-MO, CCR-KS
`23· Lexitas Legal
`· · 1608 Locust Street
`24· Kansas City, Missouri 64108
`· · 816.221.1160
`25· 800.280.3376
`
`IPR2023-00037
`Apple EX1035 Page 4
`
`
`
`·1· · · · · · IT IS HEREBY STIPULATED AND AGREED by and
`
`Page 5
`
`·2· between counsel for the Petitioner and counsel for the
`
`·3· Patent Owner that this deposition may be taken in
`
`·4· shorthand by Lauren N. Lawrence, RPR, CCR-MO, CCR-KS,
`
`·5· and afterwards transcribed into typewriting; and the
`
`·6· signature of the witness is expressly reserved.
`
`·7· · · · · · · · · ·*· · *· · *· · *· · *
`
`·8· · · · · · (Deposition commenced at 9:37 a.m.)
`
`·9· · · · · · · · ·DAVID V. ANDERSON, PH.D.,
`
`10· having been first duly sworn and examined on behalf of
`
`11· · · · · ·the Petitioner, testified as follows:
`
`12· · · · · · · · · · · · EXAMINATION
`
`13· BY MS. BAILEY:
`
`14· · · · ·Q.· ·Good morning, Dr. Anderson.· Can you --
`
`15· · · · ·A.· ·Good morning.
`
`16· · · · ·Q.· ·-- confirm for me, please, that you have
`
`17· electronic copies of the record for accessing today?
`
`18· · · · ·A.· ·Yes, I do.
`
`19· · · · ·Q.· ·Okay.· And could you please confirm that
`
`20· you don't have any notes, papers, or any other
`
`21· information regarding these matters in front of you
`
`22· other than the electronic copies of the record.
`
`23· · · · ·A.· ·I do have a list of some typos in front of
`
`24· me that I wanted to go over.
`
`25· · · · ·Q.· ·Okay.· Sure.
`
`IPR2023-00037
`Apple EX1035 Page 5
`
`
`
`Page 6
`·1· · · · ·A.· ·That's the only thing I have in front of me
`
`·2· that's not just the electronic copies.
`
`·3· · · · ·Q.· ·Sure.· Do you want to go ahead and go over
`
`·4· those typos now?
`
`·5· · · · ·A.· ·Sure.· In paragraph 140 -- I'm sorry.
`
`·6· Of -- in my 140 declaration, paragraph 4 --
`
`·7· · · · · · · (Reporter clarification.)
`
`·8· · · · ·A.· ·In my 140 declaration, paragraph 4, it says
`
`·9· "277."· That should say "140."
`
`10· · · · · · · In paragraph 12 of the same declaration, it
`
`11· says "3 years."· That should say "30 years."
`
`12· · · · · · · In paragraph 45 -- it's not a big deal, but
`
`13· there's an extra word.· "Have," h-a-v-e.
`
`14· · · · · · · And then in -- after paragraph 30, on
`
`15· page 15, the numbering restarts.· So when we've been
`
`16· discussing this, we give both the page number and the
`
`17· paragraph number because it restarts to 30 -- or I'm
`
`18· sorry to 23, which means that there are two 23s up to
`
`19· two 29s, and so we might need to be specific with
`
`20· those.
`
`21· · · · · · · And then in my 319 declaration,
`
`22· paragraph 4, "277" should say "319."
`
`23· · · · ·Q.· ·(By Ms. Bailey)· Thank you, Dr. Anderson.
`
`24· · · · · · · Was there any other further typographical
`
`25· errors you wanted to correct?
`
`IPR2023-00037
`Apple EX1035 Page 6
`
`
`
`·1· · · · ·A.· ·That's it.
`
`Page 7
`
`·2· · · · ·Q.· ·Do you understand that today's deposition
`
`·3· is covering two IPRs, one IPR for the 319 patent and
`
`·4· one IPR for the 140 patent?
`
`·5· · · · ·A.· ·Yes.
`
`·6· · · · ·Q.· ·So I'm going to ask you a number of
`
`·7· questions today.· And in an effort to avoid repetitive
`
`·8· questions that may be applicable to both IPRs, when I
`
`·9· ask a question, I'm going to try to make clear if it's
`
`10· specific to the 140 patent that I'm asking a question
`
`11· specific to the 140.· And, similarly, if it's specific
`
`12· to the 319, I'm asking a question specific for the 319
`
`13· patent.· If I don't specify, is it fair or are you
`
`14· comfortable with me not specifying and that you are
`
`15· understanding that your answer is applicable to both
`
`16· patents?
`
`17· · · · ·A.· ·Yes.
`
`18· · · · ·Q.· ·And, of course, if you have a question or
`
`19· you're unsure which patent we're talking about, please
`
`20· ask me, and I am very happy to specify.
`
`21· · · · ·A.· ·Okay.
`
`22· · · · ·Q.· ·Are you aware --
`
`23· · · · · · · MR. NOROOZI:· Just for a minute, I mean,
`
`24· I -- that raises a concern from my perspective,
`
`25· because there are different combinations of references
`
`IPR2023-00037
`Apple EX1035 Page 7
`
`
`
`Page 8
`·1· at issue, and there are different opinions and -- and
`
`·2· so forth.· So I think it raises, you know, a
`
`·3· significant risk of confusion if -- if we're talking
`
`·4· generically without specifying.· But I will just say
`
`·5· if, in any instance, there's a lack of ambiguity -- I
`
`·6· mean, there's a lack of clarity, then, you know, I'll
`
`·7· leave it to the witness and the questioning attorney
`
`·8· to sort that out between you.· I'll make my
`
`·9· objections.
`
`10· · · · · · · MS. BAILEY:· Understood.· And, Counsel, if
`
`11· you also are unclear which question or which patent I
`
`12· am asking a question for, please let me know.· Again,
`
`13· I'm happy to clarify.
`
`14· · · · · · · MR. NOROOZI:· Okay.
`
`15· · · · ·Q.· ·(By Ms. Bailey)· Dr. Anderson, are you
`
`16· aware of the legal concept in patent law regarding
`
`17· enablement of a patent?
`
`18· · · · ·A.· ·I'm not a lawyer, but I know that basically
`
`19· enablement has to do with whether the patent teaches
`
`20· how to implement or use a particular invention.
`
`21· · · · ·Q.· ·Do you think that the enablement standard
`
`22· requires a patent to lay out detail on how to make and
`
`23· use the claimed invention that would otherwise be
`
`24· known to the skilled artisan?
`
`25· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`IPR2023-00037
`Apple EX1035 Page 8
`
`
`
`Page 9
`·1· · · · ·A.· ·So I didn't understand that enablement was
`
`·2· at issue here, and so I haven't spent time preparing,
`
`·3· you know, to understand what exactly enablement means
`
`·4· in legal terms.
`
`·5· · · · ·Q.· ·(By Ms. Bailey)· Are you aware that the
`
`·6· claim priority date for the 140 and 319 patents is
`
`·7· February 4, 2002?
`
`·8· · · · · · · And if it's helpful for you, if you want to
`
`·9· turn to your declaration for the 140 patent,
`
`10· paragraph 14...
`
`11· · · · ·A.· ·Yes.· I see in paragraph 15, it says "I
`
`12· understand that the petition applies a priority date
`
`13· of February 4, 2002, for the challenged claims, and I
`
`14· apply the same date."
`
`15· · · · ·Q.· ·So if I refer to the priority date of the
`
`16· patents, can we agree that I'm referring to
`
`17· February 4, 2002?
`
`18· · · · ·A.· ·Yes.· Is that the same for the 319, though?
`
`19· Oh, it is.· I see it there.· Yes.· I'm good with that.
`
`20· · · · ·Q.· ·Could you please turn to your declaration
`
`21· for the 140 patent and read paragraph 16 to yourself?
`
`22· · · · ·A.· ·Okay.
`
`23· · · · ·Q.· ·It is -- is it your understanding that this
`
`24· paragraph is laying out the definition of a person of
`
`25· skill in the art as set forth in the petition?
`
`IPR2023-00037
`Apple EX1035 Page 9
`
`
`
`·1· · · · ·A.· ·Yes.
`
`Page 10
`
`·2· · · · ·Q.· ·Is it your understanding that this same
`
`·3· level of skill is used and applied for the 319 patent?
`
`·4· · · · ·A.· ·Yes.
`
`·5· · · · ·Q.· ·And do you understand that this level of
`
`·6· skill for a person of skill in the art includes both a
`
`·7· minimum education and a minimum years of experience?
`
`·8· · · · ·A.· ·Yes.
`
`·9· · · · ·Q.· ·Okay.· Do you dispute that this is a
`
`10· correct level of skill for the 140 and the 319
`
`11· patents?
`
`12· · · · ·A.· ·I do not.
`
`13· · · · ·Q.· ·Do you think you are a person of skill in
`
`14· the art applying this definition?
`
`15· · · · ·A.· ·Yes.
`
`16· · · · ·Q.· ·What are your qualifications that make you
`
`17· a person of skill in the art applying this definition?
`
`18· · · · ·A.· ·I have a master's degree in electrical and
`
`19· computer engineering and -- which I received in 1994.
`
`20· · · · · · · After that, I received a PhD in signal
`
`21· processing focusing on speech and audio processing,
`
`22· including speech recognition.
`
`23· · · · · · · After the PhD, I became a professor at
`
`24· Georgia Tech and have since done research and taught
`
`25· classes in speech recognition and other areas of
`
`IPR2023-00037
`Apple EX1035 Page 10
`
`
`
`·1· signal processing.
`
`Page 11
`
`·2· · · · ·Q.· ·Do you agree that this is a correct level
`
`·3· of skill for the 140 and 319 patents?
`
`·4· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`·5· · · · ·A.· ·I adopted this definition of a person of
`
`·6· ordinary skill in the art based on the petition's
`
`·7· definition.· I don't necessarily agree, but I'm --
`
`·8· it's a reasonable place to consider this from,
`
`·9· evaluate these patents from.
`
`10· · · · ·Q.· ·(By Ms. Bailey)· Why do you not necessarily
`
`11· agree with the definition?
`
`12· · · · ·A.· ·I haven't provided an alternative
`
`13· definition.· I didn't really think to.· I saw that
`
`14· this was a reasonable place to start, and since it was
`
`15· proposed by the petitioner, I adopted it.· If I were
`
`16· given time to consider, I might alter it in some way.
`
`17· I'm not sure yet.
`
`18· · · · ·Q.· ·Dr. Anderson, quite a bit of your opinions
`
`19· are based on what a skilled person would or would not
`
`20· know.· So are you stating today that you didn't really
`
`21· think to consider the level of skill?
`
`22· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`23· · · · ·A.· ·No.· That's not what I'm saying.
`
`24· · · · ·Q.· ·(By Ms. Bailey)· Could you please clarify
`
`25· then what you are saying?
`
`IPR2023-00037
`Apple EX1035 Page 11
`
`
`
`Page 12
`·1· · · · ·A.· ·I'm saying that the petitioner put forth a
`
`·2· person of ordinary skill in the art definition.· It
`
`·3· was reasonable.· Somebody with experience in speech
`
`·4· recognition.· I believe somebody with those particular
`
`·5· backgrounds could implement the patents at issue, and
`
`·6· so those are the ones that I used.· I took their
`
`·7· definition because that was where the arguments were
`
`·8· made from.· Didn't propose a different one.· That's
`
`·9· not my role.· Instead, I evaluated the obviousness and
`
`10· motivation to combine and other things from the
`
`11· definition given by the petitioner.· The main issue
`
`12· that I have -- as you said, a lot comes down to a
`
`13· person of ordinary skill in the art.
`
`14· · · · · · · The main problem I have isn't that this
`
`15· defined person of ordinary skill in the art could not
`
`16· make use of the claimed inventions in the 319 and 140
`
`17· patent.· The main issue is that the petitioner put
`
`18· forward some combinations where there are elements
`
`19· missing and would need to be filled in.· And this
`
`20· person of ordinary skill in the art would not have the
`
`21· requisite skills or understanding to do design in
`
`22· another field as required by the petitioner's
`
`23· combinations.
`
`24· · · · ·Q.· ·Do you have an opinion on what the
`
`25· appropriate level of skill should be for the 140 and
`
`IPR2023-00037
`Apple EX1035 Page 12
`
`
`
`·1· 319 patents?
`
`Page 13
`
`·2· · · · ·A.· ·As I've said, I adopted the one given, and
`
`·3· I think it's reasonable.· I'm not proposing anything
`
`·4· different.
`
`·5· · · · ·Q.· ·Do you think an engineer with a master's
`
`·6· degree in EE, CS, or CE would have been trained in
`
`·7· evaluating both the costs and benefits of a particular
`
`·8· design choice in developing a signal processing
`
`·9· architecture?
`
`10· · · · ·A.· ·Let me break that question into two parts.
`
`11· I do think that an electrical engineering student or
`
`12· CS, more particularly, in electrical engineering in
`
`13· computer engineering, would be very versed in
`
`14· evaluating design tradeoffs, in general, in whatever
`
`15· area of expertise is.· They would all have the -- be
`
`16· trained to evaluate design tradeoffs.· But in signal
`
`17· processing architecture, for example, it depends on
`
`18· what their background is.
`
`19· · · · · · · If, for example, you were in electrical
`
`20· engineering and trained in fiber optics or analog
`
`21· circuits, you would have a very different toolbox with
`
`22· which to evaluate design tradeoffs.· You would
`
`23· understand the need to, but the specifics would depend
`
`24· on the discipline.
`
`25· · · · ·Q.· ·And do you agree that the definition for
`
`IPR2023-00037
`Apple EX1035 Page 13
`
`
`
`Page 14
`·1· the skilled person provided in the petitions for the
`
`·2· 140 and 319 patents allows for two years of experience
`
`·3· after graduation?
`
`·4· · · · ·A.· ·Yes.
`
`·5· · · · ·Q.· ·Could there be a person that studied
`
`·6· high-performance computing and parallel processing in
`
`·7· their bachelor's degree, studying speech recognition
`
`·8· in their master's and then spent two years building a
`
`·9· system architecture specific to speech recognition,
`
`10· does such a person potentially exist?
`
`11· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`12· · · · ·A.· ·You're describing something that I wouldn't
`
`13· call a person of ordinary skill in the art, in general
`
`14· or specifically with this.· Someone who -- who studies
`
`15· high-performance computer architecture in their
`
`16· bachelor's degree would not have the knowledge and
`
`17· experience to design new high-performance architecture
`
`18· wholesale.· That usually comes at the graduate level.
`
`19· Furthermore, such a person would have focused on
`
`20· classes in circuit design, computer architecture, not
`
`21· in signal processing.
`
`22· · · · · · · At Georgia Tech, we have three
`
`23· undergraduate classes in signal processing for those
`
`24· desiring to take that chain, plus classes in signal
`
`25· processing applications that prepare somebody for
`
`IPR2023-00037
`Apple EX1035 Page 14
`
`
`
`Page 15
`·1· graduate study in signal processing.· If you were to
`
`·2· go straight into that, you would need to go back and
`
`·3· take the prerequisite courses to understand just the
`
`·4· signaling processing concepts.
`
`·5· · · · · · · Once that's done, they would then study
`
`·6· speech recognition.· In a master's degree, by the time
`
`·7· they got the prerequisite chain out of the way, they
`
`·8· should have already graduated.· So you're -- you're
`
`·9· giving some hypothetical superman that just doesn't
`
`10· fit the reality of a ord- -- person of ordinary skill
`
`11· in the art.
`
`12· · · · ·Q.· ·(By Ms. Bailey)· As of the priority date of
`
`13· the 140 and 319 patents, were persons working on
`
`14· speech recognition?
`
`15· · · · ·A.· ·In general, were there people working on
`
`16· speech recognition as of those priority dates?· Is
`
`17· that what you're asking?
`
`18· · · · ·Q.· ·Correct.· That's my question.
`
`19· · · · ·A.· ·Yes.
`
`20· · · · ·Q.· ·Prior to February 2002, in the field of
`
`21· speech recognition, were there methods available to
`
`22· make speech recognition more efficient.
`
`23· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`24· · · · ·A.· ·In general, yes.
`
`25· · · · ·Q.· ·(By Ms. Bailey)· Did these methods include
`
`IPR2023-00037
`Apple EX1035 Page 15
`
`
`
`·1· developments and hardware architecture for speech
`
`Page 16
`
`·2· recognition?
`
`·3· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`·4· · · · ·A.· ·Most of the work -- I would say the vast
`
`·5· majority of the work in speech recognition was on
`
`·6· software systems and the mathematics behind them.
`
`·7· There are a few groups that focus also on hardware
`
`·8· accelerators.· Most of those were cited either in the
`
`·9· specifications of the patents or in the prior art, and
`
`10· those generally involved multi-disciplinary teams or
`
`11· people who had specific training in hardware
`
`12· architectures and speech recognition applications.
`
`13· · · · ·Q.· ·(By Ms. Bailey)· As of the priority date of
`
`14· the 140 and 319 patents, was the use of parallel
`
`15· processing techniques known to improve speech
`
`16· recognition?
`
`17· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`18· · · · ·A.· ·Based on the prior art that's been
`
`19· presented, some -- actually, it wasn't known generally
`
`20· to improve speech recognition.· There were some cases
`
`21· where parallel processing was known, but there are
`
`22· also cases -- this is a very complex problem, for
`
`23· example cited by Mathew, where parallel lies
`
`24· [phonetic] in the search stage was shown to actually
`
`25· decrease the overall performance due to the
`
`IPR2023-00037
`Apple EX1035 Page 16
`
`
`
`·1· synchronization requirements and handling and
`
`Page 17
`
`·2· coordinating the data along the different modules.
`
`·3· · · · ·Q.· ·(By Ms. Bailey)· You just prefaced the
`
`·4· answer to my previous question as based on the prior
`
`·5· art that's been presented, but I am asking you, as a
`
`·6· person of skill in the art, per your earlier
`
`·7· testimony, whether you knew that adding parallel
`
`·8· processing increased the efficiency for speech
`
`·9· recognition?
`
`10· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`11· · · · ·A.· ·So -- so I'm clear, you're saying as a
`
`12· person of ordinary skill in the art, in 2002, in that
`
`13· situation, would I have known that parallel processing
`
`14· would increase the efficiency of speech recognition?
`
`15· Is that -- is that your question?
`
`16· · · · ·Q.· ·(By Ms. Bailey)· Let me rephrase my
`
`17· question so it's clear and so the record is clear.
`
`18· · · · · · · As of 2002, are you aware of research being
`
`19· performed to increase the efficiency of speech
`
`20· recognition by implementing parallel processing
`
`21· methods?
`
`22· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`23· · · · ·A.· ·So in 2002, I don't recall what I knew at
`
`24· that time, but as I'm sitting here now and looking
`
`25· back at references, citations and other things about
`
`IPR2023-00037
`Apple EX1035 Page 17
`
`
`
`·1· the prior art in 2002, there are multiple teachings
`
`Page 18
`
`·2· about parallelization.· Parallelization in speech
`
`·3· recognition is a huge topic.· It's like saying "Is
`
`·4· medication known to help a disease?"· What medication?
`
`·5· What disease?
`
`·6· · · · · · · In speech recognition, you have multiple
`
`·7· phases or multiple stages or multiple steps, and most
`
`·8· of the prior art that did -- attempted parallelization
`
`·9· with signal processing worked on parallelization, the
`
`10· operation of those steps.· For example, having feature
`
`11· extraction running in parallel with distance
`
`12· calculation or something like that.· I don't recall
`
`13· seeing anything about -- no.· I guess there is
`
`14· something.
`
`15· · · · · · · There are a variety of parallelization
`
`16· techniques.· And as I quoted from Mathew, one of the
`
`17· prior art references earlier, people were exploring
`
`18· parallelization in many aspects of the speech
`
`19· recognition problem.· Mathew specifically looked at
`
`20· parallelizing the search stage, which I assume is at
`
`21· the bottom of your question, because that's what these
`
`22· patents are associated with.
`
`23· · · · · · · Mathew found that their implementation --
`
`24· and they were experts in the area -- actually did not
`
`25· lead to an improvement in performance.· So there was
`
`IPR2023-00037
`Apple EX1035 Page 18
`
`
`
`·1· work in parallelization.· Sometimes it paid off.
`
`Page 19
`
`·2· Sometimes it didn't.· It very much depended on the
`
`·3· particular problem and the implementation.
`
`·4· · · · ·Q.· ·(By Ms. Bailey)· As of the priority date of
`
`·5· the 140 and 319 patents, was there a desire for
`
`·6· improving speech recognition systems by making them
`
`·7· more flexible?
`
`·8· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`·9· · · · ·A.· ·What do you mean by "more flexible"?
`
`10· · · · ·Q.· ·(By Ms. Bailey)· Please turn to the 140
`
`11· patent, column 4, line 6.· Feel free to read any
`
`12· portions surrounding that section that you'd like to
`
`13· read to yourself to familiarize yourself.
`
`14· · · · ·A.· ·Oh, I was reading the wrong column.· Sorry.
`
`15· · · · ·Q.· ·It would probably be helpful to you just to
`
`16· read the entire paragraph that begins on column 3,
`
`17· line 62 and extends to column 4, line 18.
`
`18· · · · ·A.· ·Thank you.
`
`19· · · · · · · Okay.
`
`20· · · · ·Q.· ·Do you see that this paragraph that you
`
`21· just read in the 140 patent mentions "flexibility" at
`
`22· least three times?
`
`23· · · · ·A.· ·Yes.
`
`24· · · · ·Q.· ·So as of the priority date of the 140 and
`
`25· 319 patents, was there a desire for improving speech
`
`IPR2023-00037
`Apple EX1035 Page 19
`
`
`
`·1· recognition systems by making them more flexible?
`
`Page 20
`
`·2· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`·3· · · · ·A.· ·So your question is very general.
`
`·4· "Flexible" could mean so many different things.· Many,
`
`·5· many different things.· Flexible in terms of
`
`·6· understanding different languages.· That would be
`
`·7· flexibility.· Flexible in terms of being easily
`
`·8· tuneable to operate under different noise situations.
`
`·9· Flexible in that it was software only, so that it
`
`10· could be implemented in a variety of different
`
`11· platforms using only a C compiler.
`
`12· · · · · · · This particular flexibility talks about
`
`13· something that I haven't seen expressed as a goal in
`
`14· any of the prior art systems.· It really hasn't come
`
`15· up because this one talks about the benefits of a
`
`16· limited segmentation of the lexical data while
`
`17· maintaining a flexibility in how the tasks are divided
`
`18· up among that group.· And since we didn't have
`
`19· anything like that in the prior art, nobody has
`
`20· expressed that that was a lack.
`
`21· · · · ·Q.· ·(By Ms. Bailey)· Do you think that it was
`
`22· known as of the priority date of the 140 patent and
`
`23· 319 patent that by having a plurality of processors in
`
`24· a group with a common memory there would be increased
`
`25· flexibility in the processing?
`
`IPR2023-00037
`Apple EX1035 Page 20
`
`
`
`·1· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`Page 21
`
`·2· · · · ·A.· ·So let's see.· I'm trying to parse your
`
`·3· question here.· If you have multiple processors in the
`
`·4· common memory, then there are more things to consider
`
`·5· how you would divide up the tasks, for example; how
`
`·6· you would schedule things; whether you're going to use
`
`·7· threading.
`
`·8· · · · · · · Generally, I remember what it was like to
`
`·9· develop multithreaded applications in -- at that time.
`
`10· And multithreading libraries and techniques were still
`
`11· under continual development.· So there is -- there are
`
`12· more design choices that could be made.· And if that's
`
`13· what you mean by "flexibility," then yes.· That's not
`
`14· the same flexibility as -- as is discussed in 4 -- in
`
`15· column 4 of the 140 patent.
`
`16· · · · ·Q.· ·(By Ms. Bailey)· As of the priority date
`
`17· for the 140 and 319 patents, was there a general
`
`18· desire to make speech recognition systems scalable to
`
`19· handle larger vocabularies?
`
`20· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`21· · · · ·A.· ·At the time of the 140 patent, very large
`
`22· vocabulary speech recognition was known.· And at the
`
`23· time, once you are at a very large vocabulary of
`
`24· speech recognition, the focus is more on improving the
`
`25· performance of the recognizer as a signal processing
`
`IPR2023-00037
`Apple EX1035 Page 21
`
`
`
`·1· task.· There is one other aspect of this.· And that
`
`Page 22
`
`·2· was at the time also some of the commercial speech
`
`·3· recognition approaches and others were focused on
`
`·4· domain-specific, large-vocabulary speech recognition.
`
`·5· · · · · · · That's something that Thelen did.· Thelen
`
`·6· solved that problem by just running many different
`
`·7· speech recognizers each specific to that domain.
`
`·8· · · · · · · But even now, there's -- we don't have the
`
`·9· concept of extremely super large speech recognition
`
`10· systems.· So the concept of a large vocabulary speech
`
`11· recognition system was known, and that was what the
`
`12· 140 patent was addressing.
`
`13· · · · ·Q.· ·(By Ms. Bailey)· Are you aware of a concept
`
`14· of reduction of practice in patent law?
`
`15· · · · · · · MR. NOROOZI:· Objection to form.
`
`16· · · · ·A.· ·I know there is such a concept.· I couldn't
`
`17· opine on what meets that.
`
`18· · · · ·Q.· ·(By Ms. Bailey)· Do you think it would have
`
`19· been difficult for a skilled person to have made the
`
`20· claimed invention of the 140 and 319 patents?
`
`21· · · · · · · MR. NOROOZI:· Objection.· Form.
`
`22· · · · ·A.· ·I haven't formed an opinion on that. I
`
`23· think, generally, that a person of ordinary skill in
`
`24· the art and speech recognition as described could make
`
`25· use of the 140 patent.· But since I don't know what
`
`IPR2023-00037
`Apple EX1035 Page 22
`
`
`
`·1· "reduction of practice" means specifically, I mean,
`
`Page 23
`
`·2· I'm not ready to give a formal opinion on it, but I
`
`·3· was satisfied that there is sufficient detail in the
`
`·4· 140 and 319 patents that it could be made use of by
`
`·5· someone of ordinary skill in the art.
`
`·6· · · · ·Q.· ·(By Ms. Bailey)· Would coordinating
`
`·7· multiple caches be something that would need to be
`
`·8· addressed when implementing the invention -- the
`
`·9· claimed inventions of the 140 and 319 patents?
`
`10· · · · ·A.· ·So the 140 and 319 specifically talk about
`
`11· how the memory structure is to work.· It gives express
`
`12· details on the different memories on the buses, on the
`
`13· fact that you will have local memories at the
`
`14· processors for control, and it says what information
`
`15· is stored where.
`
`16· · · · · · · In my declaration, I discuss the difficulty
`
`17· of various things pertaining to designing a -- a
`
`18· moving -- for example, Jiang's recognition system or
`
`19· techniques to Chen's clustered parallel processing
`
`20· system.· And coordinating multiple caches is one of
`
`21· the things that's mentioned.· It turns out that the
`
`22· 140 has solved that.· But without the benefit of the
`
`23· 140 patent, this is a question that would have to be
`
`24· addressed by anyone who was to try to move Jiang's
`
`25· system onto Chen's architecture.
`
`IPR2023-00037
`Apple EX1035 Page 23
`
`
`
`Page 24
`·1· · · · ·Q.· ·Can you point me to where in the 140 patent
`
`·2· that there's a discussion of solving the problem of
`
`·3· caches and storage that you just addressed?
`
`·4· · · · ·A.· ·Yes.· This will take me a minute because I
`
`·5· don't have my marked-up copies.
`
`·6· · · · ·Q.· ·Understood.· And feel free to word search
`
`·7· if that will make it faster for you.
`
`·8· · · · ·A.· ·So for starters -- and I'll feed you stuff
`
`·9· as I run into it, so there might be pauses in between.
`
`10· · · · · · · The 140 patent and the specification for
`
`11· the 319 is the same.· So we'll just go through it -- a
`
`12· description of the 140 patent.
`
`13· · · · · · · The 140 patent specifies what program
`
`14· data -- program and data memory is, where the language
`
`15· model memory is, where the processor is, where the
`
`16· results memory are.· It also specifies where the
`
`17· feature verdict buffer is, and it specifies connection
`
`18· for each of these to their respective processors that
`
`19· are needed.
`
`20· · · · · · · It expresses where the acoustic model
`
`21· memory is.· The acoustic model memory is specified to
`
`22· hold the acoustic model, which would be a fixed thing.
`
`23· · · · · · · Finally, it gives lexical tree processors
`
`24· connected to acoustic model memory, the results
`
`25· memory, language model processor, and the feature
`
`IPR2023-00037
`Apple EX1035 Page 24
`
`
`
`·1· verdict buffer, and so this is a different
`
`Page 25
`
`·2· architecture than Chen.· Chen, instead, does not have
`
`·3· a universally connected results memory.
`
`·4· · · · · · · Chen -- and this is particularly relevant
`
`·5· for the 319 -- does not have a universally connected
`
`·6· feature verdict buffer.· Chen, instead, envisions
`
`·7· having distributed control and data which is to be
`
`·8· shared with -- globally has to be hopped from unit to
`
`·9· unit.
`
`10· · · · · · · In such a case, the data will be updating
`
`11· in the local memories -- in the cluster memories
`
`12· accessible to all processors because that's the only
`
`13· memory he discloses having, and that's the memory that
`
`14· Schmandt has pointed to as fulfilling all the
`
`15· operations in the combination.
`
`16· · · · · · · So by having this custom architecture in
`
`17· the 140 patent, you don't have to worry about cache
`
`18· consistency with the acoustic model memory because the
`
`19· acoustic model is established at the beginning.
`
`20· · · · · · · On the other hand, when you have the
`
`21· cluster memory continually being updated as items are
`
`22· being cascaded through and it's being used for
`
`23· multiple purposes, then cache consistency becomes an
`
`24· issue.
`
`25· · · · · · · Okay.· So now let's look at other things
`
`IPR2023-00037
`Apple EX1035 Page 25
`
`
`
`·1· relevant to your question.
`
`Page 26
`
`·2· · · · ·Q.· ·And, Dr. Anderson, before you move on, for
`
`·3· clarification, could you tell me where you were
`
`·4· looking at in the 140 patent in answering that
`
`·5· question?· Was there something specific?
`
`·6· · · · ·A.· ·I started on Figure 2.
`
`·7· · · · ·Q.· ·Thank you.
`
`·8· · · · ·A.· ·So the memory structure that I've been
`
`·9· giving the overview of, based on Figure 2, the
`
`10· description starts at the bottom of column 1.· And
`
`11· through column 4, it gives an overview of the
`
`12· arrangement.
`
`13· · · · · · · And then the specifics are really starting
`
`14· more in column 5, where it talks about the various
`
`15· memories and their uses.· And even, for example, in
`
`16· column 6, it specifies specific items that are to be
`
`17· stored in the memory and how they are to be accessed
`
`18· and by whom.· And...
`
`19· · · · · · · The on-board memory for the various
`
`20· processors is described, for example -- it's
`
`21· elsewhere, but it's mentioned in column 9, about
`
`22· halfway down, around line 43.
`
`23· · · · · · · By having local -- or having on-board
`
`24· memory for the processor in addition to the shared
`
`25· memory, you then avoid cache inconsistency problems
`
`IPR2023-00037
`Apple EX1035 Page 26
`
`
`
`·1· because things that you're operating on are now
`
`Page 27
`
`·2· isolated to specific processors with their local
`
`·3· caches.
`
`·4· · · · ·Q.· ·Thank you, Dr. Anderson.· You just gave a
`
`·5· lot of information about what was in the 140
`
`·6· specification.· Could you please turn to Claim 1 of
`
`·7· the 140 patent?
`
`·8· · · · ·A.· ·Okay.
`
`·9
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
