`
`____________________________________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`____________________________________________
`
`
`APPLE INC.,
`
`Petitioner
`
`v.
`
`PARUS HOLDINGS, INC.,
`
`Patent Owner
`
`Case IPR2020-00686
`
`U.S. Patent No. 7,076,431
`____________
`
`
`PETITIONER APPLE INC.’S REPLY
`TO PATENT OWNER’S RESPONSE
`
`
`
`
`
`
`
`
`
`TABLE OF CONTENTS
`
`INTRODUCTION ......................................................................................... 1
`I.
`II. LIMITATION 1(C): “SPEAKER-INDEPENDENT SPEECH
`RECOGNITION DEVICE” ................................................................................ 2
`A. PARUS’S CLAIM CONSTRUCTION ................................................................... 2
`B. PARUS’S DEFINITION OF “VOICE PATTERN” ................................................... 3
`C. SUMMARY OF ARGUMENT ............................................................................. 3
`D. SPEECH RECOGNITION STEPS ........................................................................ 5
`E. LADD’S DISCLOSED “VOICE PATTERNS” ARE DIFFERENT THAN THE
`EXCLUDED “VOICE PATTERNS” OF PARUS’S CLAIM CONSTRUCTION ..................... 8
`1. Ladd’s Voice Patterns Are Key Words or Key Phrases ............................. 8
`2. Ladd Teaches Determining a Speech Pattern by Matching to a Grammar
`or Vocabulary ................................................................................................. 9
`3. Ladd’s Use of a Vocabulary to Detect a Speech Pattern Cannot Be the
`Same “Voice Pattern” Excluded from Parus’s Claim Construction ...............10
`F. PARUS MAKES NO SHOWING THAT LADD REQUIRES “PREDEFINED” VOICE
`PATTERNS ..........................................................................................................12
`G. MR. OCCHIOGROSSO’S OPINIONS ARE CONCLUSORY ....................................12
`H. LADD’S SPEECH RECOGNITION DEVICE DOES NOT ADAPT TO INDIVIDUAL
`USERS ................................................................................................................14
`III. LIMITATION 1(K) ......................................................................................15
`IV. MOTIVATIONS TO COMBINE ................................................................18
`V. CLAIMS 5-6 .................................................................................................24
`VI. CONCLUSION.............................................................................................25
`
`
`
`
`
`
`
`i
`
`
`
`TABLE OF AUTHORITIES
`
`
`Cases:
`
`Edwards Lifesciences LLC v. Cook Inc., 582 F.3d 1322 (Fed. Cir. 2009)................ 8
`
`In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981) ............................... 16, 23
`
`In re Merck & Co., Inc., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986) ............. 16
`
`KSR Int’l Co. v. TeleflexInc., 550 U.S. 398, 82 USPQ2d 1385 (2007) .................. 17
`
`Polygroup Ltd. MCO v. Willis Elec. Co., 780 F. App’x 880 (Fed. Cir. 2019) ........ 19
`
`Rembrandt Wireless Tech. LP v. Samsung Elect. Co. Ltd., 853 F.3d 1370
`(Fed. Cir. 2017) ...................................................................................................... 8
`
`Securenet Techs., LLC v. Icontrol Networks, Inc., IPR2016-01911, Paper 9
`(PTAB Mar. 31, 2017) .......................................................................................... 12
`
`TQ Delta, LLC v. CISCO Sys., Inc., 942 F.3d 1352 (Fed. Cir. 2019) ..................... 12
`
`United States v. Great Am. Ins. Co. of N.Y., 738 F.3d 1320 (Fed. Cir. 2013) ........ 25
`
`
`
`
`Regulations:
`37 C.F.R. § 42.6 .................................................................................................... 30
`
`37 C.F.R. § 42.6(e) ............................................................................................... 30
`
`37 C.F.R. § 42.8 .................................................................................................... 29
`
`37 C.F.R. § 42.24 .................................................................................................. 29
`
`37 C.F.R. § 42.65(a) ............................................................................................. 12
`
`
`
`
`
`
`
`
`ii
`
`
`
`
`I.
`
`INTRODUCTION
`
`Parus’s primary argument for patentability relies on a claim construction
`
`requiring the claimed speaker-independent speech recognition device to not
`
`recognize spoken words using “predefined voice patterns.” (Paper 15, Patent Owner
`
`Response, 21-24). Parus asserts Ladd’s speaker-independent speech recognition
`
`device does recognize spoken words using predefined voice patterns. Id. at 34-38.
`
`Critically missing from Parus’s Patent Owner Response (POR) is any explanation,
`
`let alone evidence or factual basis, that the “predefined voice patterns” excluded by
`
`Parus’s claim construction are the same “voice patterns” taught by Ladd. Parus’s
`
`POR did not even explain Parus’s opinion of what constitutes a “voice pattern.” It
`
`was only after deposition questioning of Parus’s declarant that Parus provided its
`
`definition of “voice pattern.” Parus’s excluded “voice pattern” is wholly different
`
`than the speech/voice patterns described in Ladd. Because Parus provides no
`
`analysis or evidence of how Ladd’s described speech/voice patterns are the same as
`
`Parus’s excluded voice patterns, Parus does not establish Ladd fails to teach the
`
`claimed speaker-independent speech recognition device.
`
`Parus’s remaining arguments do not respond to the Petition’s mapping or
`
`motivations to combine and consequently, provide no rebuttal arguments changing
`
`the Board’s initial decision.
`
`
`
`1
`
`
`
`II. LIMITATION 1(C): “SPEAKER-INDEPENDENT SPEECH
`RECOGNITION DEVICE”
`A.
`Parus contends the claimed “speaker-independent speech recognition device”
`
`Parus’s Claim Construction
`
`is not taught by Ladd when applying Parus’s proposed claim construction. (Paper
`
`15, 34-38). The District Court in the concurrent litigation previously construed the
`
`term. For purposes of this IPR, Apple submits the Court’s construction should be
`
`applied. Parus proposes a modification of the Court’s construction. Id. at 21. Below
`
`are the various constructions.
`
`District Court’s
`Construction
`
`Parus’s IPR
`Construction
`
`speech
`recognition
`device that
`recognizes
`spoken words
`without using
`predefined
`voice patterns
`
`Apple’s District
`Court
`Construction
`speech recognition
`device that does
`not adapt to
`individual users
`
`Parus’s District
`Court
`Construction
`device capable of
`recognizing spoken
`audible inputs that
`need not be trained
`to recognize the
`voice patterns of an
`individual user
`
`speech recognition
`device that
`recognizes spoken
`words without
`adapting to
`individual speakers
`or using predefined
`voice patterns
`
`(Ex. 1041, Claim Construction Order, 2; Paper 15, 34; Ex. 2012, 14-15).
`
`Parus bases its claim construction on the ’431 Patent’s statement of not using
`
`“predefined voice patterns” to recognize spoken voice commands. ’431 Patent, Ex.
`
`1001, 4:38-43. Parus argues that because the ’431 Patent describes using phonemes
`
`to recognize spoken voice commands and not using predefined voice patterns, the
`
`claimed speaker-independent speech recognition device excludes recognizing voice
`
`
`
`2
`
`
`
`patterns. (Paper 15, 21-24). Parus is forced to argue this claim construction because
`
`Ladd expressly teaches an automatic speech recognition unit that provides “speaker
`
`independent automatic speech recognition.” Ladd, 9:28-30; Paper 1, Petition, 22-23.
`
`B.
`
`Parus’s Definition of “Voice Pattern”
`
`Parus’s POR and supporting evidence provide no definition or explanation of
`
`what a “voice pattern” is as used in Parus’s claim construction. The ’431 Patent also
`
`does not provide any definition, only using the phrase twice at 4:38-43. When
`
`questioned during his deposition, Parus’s declarant, Mr. Occhiogrosso, provided a
`
`definition of “voice pattern” as “a word or utterance, and its spectral energy—
`
`typically—spectral energy as a function of time.” (Ex. 1039, Occhiogrosso Dep. Tr.,
`
`25:12-17, 25:22–26:13, 30:10-16).
`
`Summary of Argument
`
`C.
`Parus contends the speech/voice patterns discussed in Ladd are the same
`
`excluded “voice patterns” in the Parus claim construction, and therefore, Ladd
`
`allegedly does not teach the claimed speaker-independent speech recognition device.
`
`(Paper 15, 34-38). Ladd includes several teachings of determining whether a speech
`
`(or voice) pattern matches stored grammars or vocabulary and defines the
`
`speech/voice pattern as a key word or phrase. Ladd, 9:32-39, 8:23-25, 4:23-25. In
`
`contrast, Parus defines a voice pattern as the spectral energy of the user’s spoken
`
`words as a function of time. (Ex. 1039, 25:12-17, 25:22–26:13, 30:10-16). Parus
`
`
`
`3
`
`
`
`provides no evidence the speech/voice pattern in Ladd is the spectral energy as a
`
`function of time of the user’s spoken words.
`
`Ladd’s determination of voice patterns occurs at a different step in the speech
`
`recognition process than Parus’s excluded voice patterns, thereby further confirming
`
`Ladd’s voice patterns cannot be Parus’s excluded voice patterns. Per Mr.
`
`Occhiogrosso, a speech recognition device can convert the user’s spoken words into
`
`text using various voice recognition algorithms, including analyzing voice patterns.
`
`Id. at 50:21–51:8, 54:6–55:7. After converting the spoken words into text, the
`
`content of the word, e.g., what is commanded by the word, is determined by
`
`comparing the text to a recognition grammar. Id. at 66:5–67:17 (Mr. Occhiogrosso
`
`opining the ’431 Patent outputs recognition results comprising text in a data message
`
`and then compares the recognition results to the recognition grammar to identify
`
`keywords spoken by the user); ’431 Patent, 6:44-56.
`
`Ladd’s speech/voice pattern is determined at the second step of identifying
`
`the instructed command in the Ladd IVR system (i.e., identifying a key word or
`
`phrase) and subsequent to converting the speech into text. (Ex. 1040, Terveen Supp.
`
`Dec., 2-9).1 There is no evidence of record that Ladd’s voice patterns are the spectral
`
`energy as a function of time, and indeed, Ladd’s voice patterns cannot satisfy Parus’s
`
`
`1 All citations to Exs. 1003, 1040, and 2025 are to paragraph numbers.
`
`
`
`4
`
`
`
`definition because Ladd defines the voice patterns as key words or phrases
`
`determined by comparison to a recognition grammar.
`
`Speech Recognition Steps
`
`D.
`Speech recognition that determines the content of the spoken word is broadly
`
`divided into two steps: (1) converting the spoken utterance into text, i.e., words; and
`
`(2) determining the content of the recognized words, e.g., determining if words are
`
`keywords that issue a command. (Ex. 1039, 51:14–52:22). In an IVR system as
`
`described in the ’431 Patent and Ladd, the user is speaking words to issue commands
`
`to the system, e.g., instructing “What is the weather in Chicago?” Cf. ’431 Patent,
`
`6:44-56, with Ladd, 38:4-11, 40:25–41:10 (both discussing a system whereby the
`
`user can request weather information); Paper 1, 18, 22 (discussing same). If the
`
`speech recognition is being performed for purposes of, for example, dictation, where
`
`the system is not being commanded to perform an action based on the recognized
`
`speech, then the second step of determining the content of the words need not be
`
`performed. (Ex. 1040, 2-4; Ex. 1039, 40:2-22 (Mr. Occhiogrosso opining that
`
`without recognition grammars, the system is “effectively a dictation application with
`
`no imposed recognition grammar” that is simply translating speech into text)).
`
`Mr. Occhiogrosso agreed with this two-step process during his deposition. Mr.
`
`Occhiogrosso explained that in the first step of speech recognition, the user’s spoken
`
`words are converted into text. (Ex. 1039, 11:9-13, 15:4-20, 24:17–25:1, 33:11-20,
`
`
`
`5
`
`
`
`40:10-22, 51:14–52:22). Per Mr. Occhiogrosso, there are various algorithms for
`
`recognizing the speech and converting into text, including recognizing phonemes
`
`(which Parus contends is described in the ’431 Patent), analyzing voice patterns and
`
`comparing the voice patterns to predefined or reference voice patterns, and
`
`employing AI and neural networks. Id. at 25:2-6, 33:6-20, 54:6–55:7. Mr.
`
`Occhiogrosso testified that using predefined voice patterns is “one class of
`
`algorithms…that can be used to convert a speaker’s utterance into text.” Id. at 33:13-
`
`16, 43:16–44:7. Thus, per Mr. Occhiogrosso, speech recognition as described in the
`
`’431 Patent first recognizes a word by converting speech into text, and then
`
`compares the word to a recognition grammar to identify keywords. Id. at 38:5–39:2,
`
`49:14-19 (discussing recognizing keywords using the recognition grammar), 52:19-
`
`22 (opining “I think translation of the speech into text and the associated words that
`
`comprise that text is a prerequisite to identifying the keywords”).
`
`Notably, Mr. Occhiogrosso asserted that converting speech into text is
`
`“uncorrelated” with and distinct from recognizing keywords in the converted text,
`
`and the recognition grammar is “a higher order processing” that addresses what the
`
`words are and is performed after the word-to-text conversion. Id. at 50:17–51:8,
`
`31:9-11, 32:24–33:1, 51:14–52:22, 49:14-19. When asked why some words are
`
`merely “words” and some are “keywords” in the ’431 Patent, Mr. Occhiogrosso
`
`explained that the speech recognition algorithm performs the first step of converting
`
`
`
`6
`
`
`
`speech into text (which includes all words spoken by the user), but it is not until the
`
`second step that keywords based on the recognition grammar are determined:
`
`A: Okay, thank you. Yeah, the speech recognition engine—it’s a
`
`moniker, but the speech recognition algorithm generates words,
`keywords, definite articles, indefinite articles. It makes no never mind.
`It spits out words.
`
`There’s additional logic in the speech recognition engine based
`on the recognition grammar that is looking for keywords, and that is the
`logic that would isolate ‘weather’ and ‘Chicago’ from the textual
`message or the data message that the speech recognizer, per se,
`generates.
`
`(Ex. 1039, 49:5-19). Thus, per Mr. Occhiogrosso, the “recognition grammar
`
`comparison would be at the word level” and occurs after the speech is converted into
`
`text. Id. at 38:12-16.
`
`Per Parus, the ’431 Patent excludes using voice patterns to convert speech into
`
`text during the first step of speech recognition, i.e., when converting the spoken
`
`utterance into text. (Paper 15, 35-37; Ex. 1039, 29:14-23). In contrast, Ladd teaches
`
`the automatic speech recognition (ASR) unit 254 first recognizes the words from the
`
`user’s speech input and then performs the second step of determining whether the
`
`speech inputs match any key word or phrase via comparison to a stored grammar or
`
`vocabulary. Ladd, 9:31-36, 8:23-25; Ex. 1040, 4-8. Determination of a voice pattern,
`
`i.e., a key word or phrase, is performed only after the speech input is converted into
`
`
`
`7
`
`
`
`text. (Ex. 1040, 6-8, 11). Thus, Ladd’s speech/voice patterns are distinct from the
`
`“voice patterns” that the ’431 Patent allegedly excludes.
`
`E.
`
`Ladd’s Disclosed “Voice Patterns” Are Different Than the
`Excluded “Voice Patterns” of Parus’s Claim Construction
`1.
`
` Ladd’s Voice Patterns Are Key Words or Key Phrases
`
`Ladd teaches “the voice communication boards may include a voice
`
`recognition engine having a vocabulary for detecting a speech pattern (i.e., a key
`
`word or phrase).”2 Ladd, 8:23-25; Ex. 1040, 11. Dr. Terveen opines Ladd equates a
`
`speech pattern to a key word or phrase. (Ex. 1040, 12-14 (opining he understands
`
`the “i.e.” preceding “key word or phrase” in Ladd, 8:23-25 is defining a “speech
`
`pattern”); Edwards Lifesciences LLC v. Cook Inc., 582 F.3d 1322, 1334 (Fed. Cir.
`
`2009) (“[U]se of ‘i.e.’ signals an intent to define the word to which it refers.”);
`
`Rembrandt Wireless Tech. LP v. Samsung Elect. Co. Ltd., 853 F.3d 1370, 1376 (Fed.
`
`Cir. 2017) (“A patentee’s use of ‘i.e.,’ in the intrinsic record, however, is often
`
`definitional.”)). Dr. Terveen further explains the speech pattern being a key word or
`
`phrase is consistent with the remainder of the sentence at 8:23-25, as the stored
`
`
`2 During his deposition, Dr. Terveen opined that “speech pattern” and “voice
`
`pattern” were used interchangeably, and Parus relied on this opinion in its
`
`arguments. (Ex. 2024, 40:11-21; Paper 15, 37).
`
`
`
`8
`
`
`
`vocabulary is used for detecting a speech pattern, i.e., a key word or phrase, from
`
`the user’s speech inputs. Ex. 1040, 11-14.
`
`2.
`
`Ladd Teaches Determining a Speech Pattern by Matching to a
`Grammar or Vocabulary
`
`Other teachings in Ladd confirm the disclosed speech/voice pattern is a key
`
`word or phrase matched to the stored grammar or vocabulary. Ladd describes the
`
`user’s spoken words as the “speech inputs” or the “speech communications.” Ladd,
`
`9:28-30; Ex. 1040, 11. The speech inputs are compared against a “vocabulary” or
`
`“grammar” to detect key words or phrases. Ladd, 8:58-61, 10:3-20; Ex. 1040, 11-
`
`14. Parus relies on Ladd’s disclosure at 9:28-44. (Paper 15, 36). Dr. Terveen explains
`
`that
`
`if
`
`the user spoke a word or key word or phrase matching
`
`the
`
`grammar/vocabulary, then an output signal associated with the selected key word or
`
`phrase is sent. (Ex. 1040, 14-15). Matching of a spoken key word or phrase, i.e., a
`
`speech pattern, with a grammar/vocabulary results in the ASR “identif[ying] a
`
`selected speech of the speech inputs.” Id., citing Ladd, 9:36-38; Paper 1, 32-36
`
`(Petition mapping Ladd’s detection of the speech pattern for determining the issued
`
`command to the claimed recognition grammar and speech command comprising an
`
`information request of Limitations 1(f)-1(h) of the ’431 Patent); Ex. 1003, 105, 110-
`
`111.
`
`The “selected speech pattern” is the speech pattern, i.e., the key word or
`
`phrase, selected to be spoken by the user. (Ex. 1040, 15). Ladd repeatedly discusses
`
`
`
`9
`
`
`
`the user making selections responsive to a prompt. (Ex. 1040, 15, citing Ladd, 17:17-
`
`27, 23:40-44, Claim 8). As one example, Ladd describes a user being prompted for
`
`a type of soda (e.g., Pepsi or Coke), and the user selecting the desired soda. Ladd,
`
`23:40-44. The selected user input is the user’s selection in response to the prompt,
`
`i.e., selecting Coke. Ladd, Claim 8. When the user input includes a speech pattern,
`
`i.e., a key word or phrase, matching a grammar/vocabulary, then the selected
`
`keyword, e.g., Coke, is output, such that the ASR unit “sends an output signal to
`
`implement the specific function associated with the recognized voice pattern.” Ladd,
`
`9:36-39. Per Dr. Terveen, the selected voice pattern, i.e., the key word or phrase
`
`matching the grammar/vocabulary, instructs a selection by the user. (Ex. 1040, 15).
`
`Ladd includes other discussions of a speech/voice pattern at 4:15-18 and 6:50-
`
`57. Dr. Terveen notes each disclosure is consistent with Ladd’s express statement
`
`that a speech/voice pattern is a key word or phrase. (Ex. 1040, 15-24). Importantly,
`
`none of the disclosures would have reasonably indicated to a POSITA that Ladd uses
`
`the term “voice pattern” (or “speech pattern”) to mean the spectral energy as a
`
`function of time. Id.
`
`3.
`
`Ladd’s Use of a Vocabulary to Detect a Speech Pattern Cannot
`Be the Same “Voice Pattern” Excluded from Parus’s Claim
`Construction
`
`Even if Ladd did not expressly disclose that “speech pattern” is a key word or
`
`phrase (thereby failing to meet Parus’s excluded “voice patterns”), Ladd’s disclosed
`
`
`
`10
`
`
`
`speech patterns cannot satisfy Parus’s excluded “voice patterns.” Ladd teaches “a
`
`vocabulary for detecting a speech pattern.” Ladd, 8:23-25. Ladd generally equates a
`
`vocabulary with a recognition grammar. Ladd, 6:25-29; Ex. 1040, 10 (Dr. Terveen
`
`opining Ladd generally equates a grammar with a vocabulary given Ladd’s repeated
`
`reference to “grammars (i.e., vocabulary)”). Per Mr. Occhiogrosso, using a
`
`vocabulary (or grammar) to detect speech occurs after the spoken words are
`
`converted into text. (Ex. 1039, 49:9-19, 50:17–51:6). Because the “voice patterns”
`
`defined by Parus are analyzed to convert words into text, and because Ladd’s speech
`
`patterns are detected with a vocabular/grammar, Ladd’s speech patterns cannot be
`
`the same as the excluded “voice patterns.”
`
`Mr. Occhiogrosso also stated twice during his deposition that grammars have
`
`nothing to do with [predefined] voice patterns and are “uncorrelated” to voice
`
`patterns. Id. at 31:9-11, 32:24–33:1. Logically, if Ladd detects the speech pattern
`
`using a vocabulary, per 8:23-25, and if recognition grammars (which Ladd equates
`
`to a vocabulary, as discussed above) are uncorrelated with and do not have anything
`
`to do with the excluded voice patterns in the claim construction, then Ladd’s
`
`speech/voice patterns are different than the excluded voice patterns of the claim
`
`construction. (Ex. 1040, 10).
`
`In sum, because Ladd expressly teaches a speaker-independent speech
`
`recognition device, and because Ladd’s determined “voice patterns” are materially
`
`
`
`11
`
`
`
`different than the “predefined voice patterns” Parus contends are excluded by its
`
`claim construction, Ladd teaches Claim 1(c). Applying the Court’s construction,
`
`Ladd teaches a speaker-independent speech recognition device.
`
`F.
`
`Parus Makes No Showing that Ladd Requires “Predefined” Voice
`Patterns
`
`Parus makes no showing Ladd’s voice patterns are the excluded “predefined”
`
`voice patterns. (Ex. 1039, 30:10-16 (Mr. Occhiogrosso defining a “predefined” voice
`
`pattern)). Parus’s entire discussion focuses on Ladd allegedly teaching a “voice
`
`pattern,” but there is no discussion of whether the voice pattern is “predefined” or
`
`not. For this reason alone, Parus has not established that Ladd fails to teach a
`
`speaker-independent speech recognition device.
`
`G. Mr. Occhiogrosso’s Opinions Are Conclusory
`Mr. Occhiogrosso’s opinions are extremely deficient,
`
`lacking any
`
`explanation, let alone factual basis, and mostly repeating the POR’s arguments. His
`
`opinions should be given no weight. Securenet Techs., LLC v. Icontrol Networks,
`
`Inc., IPR2016-01911, Paper 9, 11, 14–15 (PTAB Mar. 31, 2017) (expert testimony
`
`that merely repeats the petition’s deficient assertions without adding more
`
`explanation or evidence is likewise deficient); TQ Delta, LLC v. CISCO Sys., Inc.,
`
`942 F.3d 1352, 1359 (Fed. Cir. 2019) (conclusory expert testimony is inadequate to
`
`support a substantial evidence review); 37 C.F.R. § 42.65(a).
`
`
`
`12
`
`
`
`Every paragraph of Mr. Occhiogrosso’s Declaration discussing why Ladd
`
`allegedly does not teach the claimed speaker-independent speech recognition device
`
`is deficient. (Paper 15, 35). Beginning with Ex. 2025, ¶ 81, Mr. Occhiogrosso’s
`
`opinions repeat the POR text at page 35 and provide no additional explanation.
`
`Paragraph 82 of Mr. Occhiogrosso’s Declaration repeats back claim construction
`
`arguments and therefore, does not otherwise expand on why Ladd allegedly does not
`
`disclose the speaker-independent speech recognition device. Paragraph 83 is a
`
`conclusory statement that Ladd does not teach the claimed device. Paragraph 84
`
`repeats some of Ladd’s teachings and again repeats the ’431 Patent’s disclosure and
`
`the POR’s argument but also does not offer any explanation for why Ladd’s speech
`
`patterns are allegedly the same as the excluded voice patterns. Paragraph 85 includes
`
`a “teach away” assertion without any explanation for why Ladd discredits,
`
`discourages, or criticizes “from the type of speech recognition required by the ‘431
`
`Patent” and again repeats back POR arguments. (Ex. 2025, 85; Paper 15, 37-38).
`
`Paragraph 85 restates Ladd’s teachings but then segues to ¶ 86, which is a conclusory
`
`sentence without explanation or factual basis. Paragraph 87 refers to speech or voice
`
`patterns being “traditionally associated” with identifying an individual speaker but
`
`not necessarily the words spoken. This paragraph is confusing, as it contradicts Mr.
`
`Occhiogrosso’s deposition testimony that voice patterns, as used in Parus’s claim
`
`construction and the ’431 Patent, are used to convert speech into text. There is also
`
`
`
`13
`
`
`
`no support or factual basis for the provided opinions. Paragraph 88 quotes Ladd’s
`
`teachings. Paragraph 89 complains Ladd allegedly does not teach certain unclaimed
`
`(and undisclosed) features of the ’431 Patent regarding “large vocabularies”
`
`(unclaimed and undisclosed in the ’431 Patent) and using phonemes (unclaimed).
`
`Finally, ¶ 90 summarizes the opinions but again, without any explanation or support.
`
`Exhibit 2025, ¶¶ 81-90 should be given no weight as conclusory and lacking factual
`
`basis.
`
`H. Ladd’s Speech Recognition Device Does Not Adapt to Individual
`Users
`
`Parus contends the full District Court construction should not be adopted.
`
`(Paper 15, 21-24). Apple disagrees and proposes adopting the Court’s construction
`
`for this IPR. Applying the Court’s construction, Ladd also teaches its disclosed
`
`speaker-independent speech recognition device does not adapt to individual users,
`
`as required by the Court’s construction. Ladd teaches an ASR unit that “provides
`
`speaker
`
`independent automatic speech
`
`recognition of speech
`
`inputs or
`
`communications from the user,” and a POSITA would have understood Ladd’s
`
`speech recognition device does not, in some embodiments, adapt to individual users.
`
`(Paper 1, 22-23; Ex. 1003, 90).
`
`Parus also agrees that Ladd is “an example” of a “speaker independent”
`
`system that does not “require training by each individual….” (Paper 15, 8-9).
`
`
`
`14
`
`
`
`III. LIMITATION 1(K)
`Parus misrepresents the Petition by arguing Goedken does not disclose
`
`sequentially accessing pre-selected web sites (Paper 15, 39) and Kurosawa does not
`
`disclose accessing pre-selected websites until the requested information is retrieved
`
`(Paper 15, 40). This argument does not respond to the Petition’s mapping. As
`
`mapped in the Petition and acknowledged by the Board at Institution, Kurosawa is
`
`relied upon for teaching sequentially accessing pre-selected web sites. (Paper 1, 39-
`
`40; Paper 9, 37-38). Goedken is relied on for teaching the procedure recited in
`
`Limitation 1(k) of accessing a first file and if the information is not found, then
`
`sequentially accessing a plurality of files until the information is retrieved. (Paper 1,
`
`43-44). Thus, Ladd as modified by Kurosawa teaches sequentially accessing a
`
`plurality of websites, and Ladd as further modified by Goedken teaches the claimed
`
`procedure in Limitation 1(k). (Paper 1, 44).
`
`Parus admits elsewhere to correctly understanding the Petition’s mapping.
`
`(Paper 15, 24-25 (“Apple relies on the Kurosawa reference (and only Kurosawa) as
`
`teaching the use of multiple web sites.”); Ex. 2025, 95 (“Petitioner and Dr. Terveen
`
`rely on Kurosawa for sequentially accessing websites, and then Goedken for short
`
`circuiting the sequential access if the requested information has been found or all of
`
`the websites have been accessed); id. at 114 (“I understand that Petitioner and Dr.
`
`Terveen are using Goedken for its algorithm of searching database files…”)).
`
`
`
`15
`
`
`
`Notably, Parus does not dispute Kurosawa teaches sequentially accessing a plurality
`
`of websites and Goedken teaches the search procedure recited in Limitation 1(k).
`
`Because Parus cannot establish the combination of Kurosawa and Goedken fails to
`
`teach Limitation 1(k), Parus resorts to improperly attacking each of the references
`
`individually. In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck
`
`& Co., Inc., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
`
`Parus also makes other unsupported statements that do not address the
`
`Petition’s mapping. For example, in discussing Goedken, Parus argues “sequential
`
`access of websites is very different than sequential access of a database.” (Paper 15,
`
`39, citing Ex. 2025, 94). Patent Owner’s and Mr. Occhiogrosso’s statements are
`
`conclusory without any factual basis. There is no explanation, for example, as to
`
`why the process of searching a database (controlled by an administrator) for
`
`information as compared to the process of searching websites for information is
`
`relevant to the claims. Patent Owner argues a difference exists, but does not explain
`
`why or how the difference is relevant to the claims of the ’431 Patent. Dr. Terveen
`
`opines he disagrees with Mr. Occhiogrosso’s opinion that the processes of searching
`
`a website versus a database are “fundamentally different.” (Ex. 1040, 26, discussing
`
`Ex. 2025, 94-95). Per Dr. Terveen, both a website and a database are information
`
`sources, and the frequency with which they are updated does not affect whether or
`
`
`
`16
`
`
`
`not they have information to be retrieved when the search for the information is
`
`performed anew in response to a request. Id.
`
`Patent Owner also argues, again without support, that “Kurosawa’s search
`
`engine is not designed or intended for use in a voice activation browser” and that a
`
`POSITA would not have been motivated to modify Kurosawa. (Paper 15, 40-41). It
`
`is well-established that work in one field of endeavor may prompt variations of it for
`
`use in either a different one based on design incentives, such as those of speed,
`
`availability, and efficiency articulated in detail in the Petition and accompanying
`
`Declaration. (Paper 1, 30-32, 44-47; KSR Int’l Co. v. TeleflexInc., 550 U.S. 398, 82
`
`USPQ2d 1385 (2007)). Regardless, Parus’s argument again does not respond to the
`
`Petition’s mapping. Kurosawa teaches “sequential access” of websites, which Parus
`
`does not dispute. Kurosawa, [0015], [0030]; Paper 1, 39-40. The Petition proposed
`
`modifying Ladd’s IVR to sequentially access a plurality of websites, as taught by
`
`Kurosawa. (Paper 1, 39-47). The Petition did not rely on Kurosawa’s search engine
`
`nor propose modifying Kurosawa (as asserted by Parus at Paper 15, bottom of page
`
`40). Therefore, Parus’s complaint that Kurosawa does not teach “stop[ping] once the
`
`request information is retrieved” fails to respond to the Petition’s mapping. (Paper
`
`15, 40).
`
`At Institution, the Board agreed Ladd in view of Kurosawa and Goedken
`
`teaches Limitation 1(k). (Paper 9, 37-38). Parus does not provide any additional
`
`
`
`17
`
`
`
`evidence to alter this finding. Because Parus’s arguments do not respond to the
`
`Petition’s mapping, Parus has waived any argument that the combination fails to
`
`teach Limitation 1(k).
`
`IV. MOTIVATIONS TO COMBINE
`
`Parus provides various arguments regarding motivations to combine, many
`
`without explaining why the argument defeats the motivations and with conclusory
`
`opinions. Apple provided extensive discussion of motivations for combining Ladd,
`
`Kurosawa, and Goedken in the interest of speed, convenience, and robustness.
`
`(Paper 1, 30-32, 34, 44-47, citing Ex. 1003, 102-103, 108, 113, 121-122). At
`
`institution, the Board found a skilled artisan would have had reason to combine these
`
`references, that Goedken does not teach away from a combination of Ladd and
`
`Kurosawa, and that the proposed mapping would not have caused delays inhibiting
`
`the combination. (Paper 9, 39-42).
`
`
`
`Parus first contends the motivation to combine the references “offers no
`
`motivation from the teachings of the asserted references” and thereby engages in
`
`hindsight. (Paper 15, 41-42, 47). This is the extent of Parus’s “hindsight” argument.
`
`There is no explanation why the motivation to combine engages in hindsight other
`
`than “there is no motivation in any of the references to combine them.” Id. at 43.
`
`Parus cites no case law supporting its argument that when motivation to combine
`
`comes from outside the references, the combination engages in hindsight. A
`
`
`
`18
`
`
`
`motivation to combine the references does not need to come from the references
`
`themselves. Polygroup Ltd. MCO v. Willis Elec. Co., 780 F. App’x 880, 884 (Fed.
`
`Cir. 2019) (“[E]vidence of a motivation to combine need not be found in the prior
`
`art references themselves….”).
`
`Parus similarly complains that Ladd, Kurosawa, and Goedken “do not provide
`
`any motivation to search a plurality of websites for information to be retrieved in
`
`order to provide the quickest,