Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 1 of 61
`
`IN THE UNITED STATES DISTRICT COURT
`FOR THE WESTERN DISTRICT OF TEXAS
`WACO DIVISION
`
`ZENTIAN LTD.,
`
`Plaintiff,
`
`
`
`v.
`
`APPLE INC.
`
`Defendant.
`
`C.A. No. 6:22-cv-122
`
`JURY TRIAL DEMANDED
`
`COMPLAINT FOR PATENT INFRINGEMENT
`
`Plaintiff Zentian Ltd. (“Zentian”) demands a trial by jury on all issues so triable and, for
`
`its complaint against Defendant Apple Inc. (“Apple”) alleges as follows:
`
`INTRODUCTION
`
`Apple introduced the “Hey Siri” feature in September 2014.
`
`“Hey Siri” allows users of Apple devices, including iPhone, iPad, Apple Watch,
`
`1.
`
`2.
`
`and Mac personal computers, among others, to engage Apple’s virtual assistant by using only their
`
`voice and without physically handling their device. Apple’s virtual assistant does so by
`
`recognizing sounds uttered by the device user and converting those acoustic waveforms into
`
`information that can be mapped into actions by the operating system, such as searching the internet,
`
`playing a song, making to-do lists, setting alarms, creating a calendar invite, or any number of
`
`other tasks. This process is known as automatic speech recognition or simply speech recognition.
`
`3.
`
`Apple’s implementation of speech recognition that enables users to wake Apple’s
`
`virtual assistant using the wake phrase “Hey Siri” and/or ask “Siri” questions hands free was
`
`invented and patented by Zentian.
`
`IPR2023-00034
`Apple EX1014 Page 1
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 2 of 61
`
`4.
`
`Zentian is a technology innovator specializing in hardware designs for speech
`
`recognition circuits and methods. Zentian was founded in 2000 by Mark Catchpole using funds
`
`from the U.K. government’s “Smart Award” program—a program that provides funding for
`
`innovative projects with the potential for significant growth.
`
`5.
`
`In the early 2000s, conventional speech recognition systems were ill-equipped to
`
`perform a feature like “Hey Siri” on a mobile electronic device in a fast and power-efficient
`
`manner, let alone facilitate the back-and-forth conversational interactions that Apple users have
`
`with Siri today. In large part, this shortcoming existed because systems in the art focused on
`
`improving algorithms and the models used by speech recognition systems to achieve higher word
`
`recognition accuracy and other goals. Such implementation focused on incremental improvements
`
`in software that ran on power and memory intensive computers or servers. Another contributing
`
`factor was the fact that mobile devices like Apple’s iPhone or Apple Watch are typically used in
`
`noisier environments, such as on public transportation or near a busy road, requiring a more
`
`complex model and more intensive computation to obtain adequate results from the speech
`
`recognition system.
`
`6.
`
`Zentian, however, focused on speech recognition hardware, including specialized
`
`chip designs to enable speech recognition in cost, space, and power constrained environments like
`
`mobile electronic devices. Zentian’s vision was to study speech recognition systems and to design
`
`silicon chips that would accelerate the speech recognition process by off-loading all or part of that
`
`processing from software running on a CPU to a bespoke silicon system. This approach differed
`
`from that of leading speech recognition companies that were focused on incrementally improving
`
`software implementations of speech recognition. Zentian focused on how to design better digital
`
`electric circuits (e.g., better hardware).
`
`-2-
`
`IPR2023-00034
`Apple EX1014 Page 2
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 3 of 61
`
`7.
`
`Critically, speech recognition systems can be divided into several discrete
`
`processes. First, an audio front-end generates processed speech parameters (referred to as feature
`
`or spectral vectors) that describe the characteristics of an audio signal. Second, a calculating means
`
`or circuit compares the similarity between feature vectors and a model of predetermined acoustic
`
`states. This stage is often referred to as “distance calculations” because the output is a probability
`
`of how similar (or different) a feature vector is from a stored acoustic state. And third, a search
`
`process attempts to find the most likely set of words spoken from a given vocabulary based on the
`
`distance calculations.
`
`8.
`
`Among the prototypes Zentian developed was a hardware accelerator for the
`
`acoustic stage (audio front end and distance calculation) of the speech recognition process that
`
`worked alongside a software implementation of the search stage running on a conventional
`
`processor. The accelerator demonstrated a two-times speedup of the speech recognition software
`
`compared to a conventional speech recognition software-only solution. Moreover, the prototype
`
`proved that segregating the speech recognition process onto bespoke hardware components could
`
`help solve the shortcomings of conventional systems and enable more advanced features like
`
`Apple’s “Hey Siri.”
`
`9.
`
`In the 2000s, Zentian had a number of conversations about its speech recognition
`
`accelerator with leading speech recognition companies, such as Nuance Communications Inc. and
`
`IBM, as well as cellular carriers, such as Hutchison 3G UK Limited. But each speech recognition
`
`company expressed a desire to focus on software-only solutions and an unwillingness to divert
`
`resources to “longer-term” ideas for speech recognition systems. At the 2006 GSM World
`
`Congress in Barcelona, a senior executive at Qualcomm, one of the preeminent suppliers of silicon
`
`chips for mobile electronic devices, tellingly stated that speech recognition was “not even in my
`
`-3-
`
`IPR2023-00034
`Apple EX1014 Page 3
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 4 of 61
`
`top ten wish list.” Put simply, the industry was not focused on hardware-centric solutions for
`
`speech recognition. Zentian was.
`
`10.
`
`Zentian’s work resulted in several U.S. patents, including U.S. Patent Numbers
`
`7,587,319; 7,979,277; 10,971,140; 10,062,377; and 10,839,789 (collectively, the “Asserted
`
`Patents”). These patents have now been widely cited on patents filed by a host of companies that
`
`provide speech recognition technology or end-user products that utilize speech recognition
`
`technology, including Samsung, Google, Amazon, Nuance Communications Inc., Microsoft, and
`
`Voci Technologies Inc.
`
`11.
`
`Today, speech recognition capabilities are in all kinds of products from phones to
`
`cars to smart speakers. Apple’s implementation of speech recognition technology in its products
`
`“on-device” involves a combination of general-purpose processors combined with hardware
`
`accelerators—the very speech recognition circuit and/or system Zentian invented and claimed in
`
`the Asserted Patents.
`
`12.
`
`Although Zentian has met with Apple on more than one occasion to discuss its
`
`technology, Apple has no license to use Zentian’s patented inventions.
`
`13.
`
`Nevertheless, Apple’s infringing on-device speech recognition features have, on
`
`information and belief, generated substantial sales of Apple products. For example, in the 12
`
`months ending on September 25, 2021, Apple reported over $191 billion in revenue for sales of
`
`iPhone, over $35 billion in revenue for sales of Mac computers, over $31 billion in revenue for
`
`sales of iPad, and over $38 billion in revenue for sales of wearables, Home and Accessories.1 On
`
`average, end users make multiple requests for Siri per day through “Hey Siri.” In turn, Siri and the
`
`
`1 Apple Inc., Condensed Consolidated Statements of Operations, Q4, Fiscal Year 2021,
`https://www.apple.com/newsroom/pdfs/FY21_Q4_Consolidated_Financial_Statements.pdf.
`
`-4-
`
`IPR2023-00034
`Apple EX1014 Page 4
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 5 of 61
`
`fluidity of the speech recognition processes that are behind it (facilitated by the hardware
`
`architecture on which it runs) have become a material driver of user demand for Apple products.
`
`14.
`
`Apple has infringed and is still infringing, directly and indirectly, one or more
`
`claims of the Asserted Patents by making, using, selling, and offering to sell in the United States,
`
`including in this District, and importing into the United States, products that implement the
`
`Asserted Patents’ speech recognition technology. Examples of potentially infringing products
`
`include Apple’s iPhone, iPad, Watch, AirPods, HomePod, HomePod Mini, Apple TV set top box,
`
`and Mac computers (including the Mac Pro, which is manufactured in Austin, Texas), in their
`
`various models and iterations (the “Accused Products”).2
`
`15.
`
`This list of Accused Products is non-limiting and based on information currently
`
`available to Zentian. Zentian reserves the right to modify the list of Accused Products as discovery
`
`progresses, including as new products are released during the pendency of this case.
`
`THE PARTIES
`
`16.
`
`Zentian is a private corporation incorporated and registered in the United Kingdom,
`
`with its registered address at 27 Queens Close, St. Ives, Cambridgeshire, England, PE37 5QD.
`
`17.
`
`18.
`
`Zentian is the owner of all rights, title, and interest in the Asserted Patents.
`
`On information and belief, Defendant Apple Inc. is a California corporation with a
`
`principal place of business at 1 Apple Park Way, Cupertino, California. Apple designs,
`
`manufactures, and sells throughout the world a wide range of products, including mobile devices
`
`and computers that incorporate Zentian’s patented technologies.
`
`
`2 See, e.g. Devices that support “Hey Siri” https://support.apple.com/en-us/HT209014. In addition
`to recognition of the wake phrase “Hey Siri,” other potentially infringing features include dictation
`of text messages and notes with integrated applications and on device speech recognition for Siri
`user requests.
`
`-5-
`
`IPR2023-00034
`Apple EX1014 Page 5
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 6 of 61
`
`JURISDICTION AND VENUE
`
`19.
`
`This is an action for patent infringement arising under the patent laws of the United
`
`States of America, Title 35 of the United States Code.
`
`20.
`
`This Court has subject matter jurisdiction over the matters asserted in this
`
`Complaint under 28 U.S.C. §§ 1331 and 1338(a) and 35 U.S.C. § 271 et seq.
`
`21.
`
`This Court’s exercise of personal jurisdiction over Apple complies with both the
`
`Texas long-arm statute and the Due Process Clause of the Fourteenth Amendment. Apple has
`
`purposely directed its activities toward Texas and has availed itself of the privileges of conducting
`
`activities in this state.
`
`22.
`
`Apple conducts continuous and systematic business in this District, including by
`
`selling infringing products to residents of this District, by soliciting business from residents in this
`
`District, and on information and belief, by manufacturing certain Accused Products in this District.
`
`23.
`
`Apple also maintains offices in Austin, Texas, and markets, offers, manufactures,
`
`and distributes the Accused Products throughout the United States, including in the state of Texas
`
`and in this judicial District.
`
`24.
`
`25.
`
`26.
`
`Apple’s Austin offices are located at 12545 Riata Vista Cir, Austin TX 78727.
`
`Venue is proper under 28 U.S.C. § 1391(b) and (c) and 28 U.S.C. § 1400(b).
`
`Apple has committed acts of infringement in this District and has a regular and
`
`established place of business here, including offices in Austin, Texas, where it maintains a
`
`workforce of more than 7,000 employees.3
`
`27.
`
`Apple has maintained a physical presence in this District for years, opening its first
`
`offices in Austin in the 1990s. And Apple continues to grow its footprint in this District with a
`
`
`in Austin, Press Release, (Nov. 20, 2019) https://www.ap-
`3 See Apple Expands
`ple.com/newsroom/2019/11/apple-expands-in-austin/.
`
`-6-
`
`IPR2023-00034
`Apple EX1014 Page 6
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 7 of 61
`
`planned $1 billion, 3-million-square-foot, 133-acre campus that will house up to 15,000
`
`employees.4 The Apple “Austin campus” will include a 192-room hotel,5 and provide a wide
`
`variety of engineering, research and development, operations, finance, sales and customer support
`
`jobs.6 On information and belief, Apple also owns significant real estate in and around Austin,
`
`Texas, including for example, a 217,000 square-foot building at 320 S. Capital of Texas Hwy.7
`
`28.
`
`Apple’s new Austin campus is responsible for manufacturing the Mac Pro, one of
`
`the Accused Products in this case. The Mac Pro and other Accused Products also contain
`
`components from suppliers across the state of Texas, including suppliers in this District. According
`
`to Apple’s Chief Executive Officer, Tim Cook, “[b]uilding the Mac Pro, Apple’s most powerful
`
`device ever, in Austin is both a point of pride and . . . deepening our close bond with the city and
`
`the talented and diverse workforce that calls it home.”8
`
`29.
`
`Apple also employs individuals in this District who, on information and belief,
`
`contribute to Apple’s design and sales of infringing products. For example, and without limitation,
`
`Apple employs over 100 individuals in the Austin metropolitan area on the “Siri Team” and/or
`
`with the title of SoC engineer.9
`
`
`
`4 Id.
`5 Juli Clover, Apple Adding 192-Room Hotel to Upcoming Austin Campus, MacRumors (May 20,
`2020) https://www.macrumors.com/2020/05/20/apple-austin-campus-hotel/.
`6 Alex Caprariello, Billy Gates, Apple’s Billion-Dollar Austin Campus Nearly Finished, Move-in
`Date Set, KXAN (Apr. 26, 2021), https://www.kxan.com/news/business/apples-1b-austin-cam-
`pus-nearly-finished-move-in-set-for-2022/.
`7 Parimal Rohit, Apple Buys Austin Office Building on Capital of Texas Hwy, Austin Business
`Journal (Aug. 25, 2021), https://www.bizjournals.com/austin/news/2021/08/25/apple-buys-capi-
`tal-ridge-austin.html.
`8 Id.; see also Apple Expands in Austin, Press Release (Nov. 20, 2019) https://www.ap-
`ple.com/newsroom/2019/11/apple-expands-in-austin/.
`9 See, e.g. Rubinia Leal Cavazos (“Team Lead, Siri Team at Apple”), LinkedIn profile available
`online at https://www.linkedin.com/in/rubinialeal/; Nicole Mohrmann (“Team Lead (Siri) at
`Apple”), LinkedIn profile available online at https://www.linkedin.com/in/nicole-mohrmann-
`
`
`-7-
`
`IPR2023-00034
`Apple EX1014 Page 7
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 8 of 61
`
`30.
`
`Apple has also sought to recruit individuals for employment within this District
`
`through its website, LinkedIn, and other websites. This includes recent postings seeking to fill
`
`hundreds of full-time positions for SoC engineers and/or Artificial Intelligence and Machine
`
`Learning Annotation Analyst on the Siri Team.10 Most of these positions require an “On-site”
`
`presence in this District.11
`
`31.
`
`Moreover, as described below, many of the Accused Products implement Apple’s
`
`infringing speech recognition technology on Intel chips. Intel is a designer and manufacturer of
`
`various processors and System on Chips. Intel is headquartered in Austin, Texas, where it employs
`
`close to 2,000 people.12 In pertinent part, this means that on information and belief, components
`
`central to Apple’s acts of infringement were designed and developed in this District, by individuals
`
`who reside and work here.
`
`32.
`
`On information and belief, critical documents regarding the Accused Products are
`
`also located in this District given the presence of Intel and Apple employees that live in the District
`
`and work on Apple’s infringing speech recognition technology.
`
`
`966ab771/; Mortuza Lokhandwala (“SOC Design Verification Engineer at Apple), LinkedIn
`profile available online at https://www.linkedin.com/in/murtuza-lokhandwala-a8b53785/; Vu Le
`(“SoC Design Engineer
`at Apple”), LinkedIn
`profile
`available
`online
`at
`https://www.linkedin.com/in/vutle/; Neil Panchal (“SoC Physical Design Manager at Apple”),
`LinkedIn profile available online at https://www.linkedin.com/in/neil-panchal-194a783/.
`10 See, e.g. Apple LinkedIn Job postings, https://www.linkedin.com/jobs/search/?current-
`JobId=2848276684&geoId=102748797&keywords=siri%20apple&location=Texas%2C%20Unit
`ed%20States; see also Careers at Apple: Find your perfect role (search “Austin, TX” in Location
`field and “SoC Architect” and/or “Siri”
`in search
`field), https://jobs.apple.com/en-
`us/search?search=SoC%20Architect&sort=relevance&location=austin-AST.
`11 Id.
`12 See Innovating and Investing in Texas, https://www.intel.com/content/www/us/en/corporate-
`responsibility/intel-in-texas.html.
`
`-8-
`
`IPR2023-00034
`Apple EX1014 Page 8
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 9 of 61
`
`ZENTIAN BREAKS NEW GROUND
`
`33.
`
`Mark Catchpole, Zentian’s founder and one of the named inventors on the Asserted
`
`Patents, holds a B.Sc. degree in Physics from the University of Edinburgh and a M.Sc. degree in
`
`Microelectronics Systems Design from Brunel University in London. After initial stints teaching
`
`physics, Catchpole joined Phoenix VLSI and later PixelFusion as an electronics engineer working
`
`on Application Specific Integrated Circuit (ASIC) and Field Programmable Gate Array (FPGA)
`
`designs. ASICs and FPGAs are two forms of integrated circuits or chips that allow for
`
`customization of hardware after manufacturing.
`
`34.
`
`Most of the work by Zentian was—and still is—unpaid, as funds proved difficult
`
`to raise during the early 2000s recession following the dot-com crash. Engineers worked in
`
`exchange for share options in the company, investing their own sweat equity into the potential of
`
`a hardware system-on-chip solution for speech recognition. During this time, Zentian was able to
`
`build a partial prototype on a programmable hardware emulation platform.
`
`35.
`
`Guy Larri joined the Zentian team in 2004 as a technical contributor. Larri holds a
`
`B.Sc. degree in Mathematics and Computer Science, as well as an engineering degree in Electrical
`
`Engineering from the University of Sydney. Larri has worked as a VLSI (very large-scale
`
`integration) chip design engineer, designing audio processor chips, cache, memory controllers, and
`
`other microprocessors.
`
`36.
`
`After Larri joined, Zentian’s efforts shifted to the creation of a hardware accelerator
`
`for the acoustic stage of the speech recognition process that would work alongside a separate
`
`software implementation of the search stage (presumably running on a conventional mobile
`
`processor). Once the design specifications were created, the Zentian acoustic accelerator was
`
`loaded onto a programmable gate array (a type of integrated circuit designed to be customized
`
`-9-
`
`IPR2023-00034
`Apple EX1014 Page 9
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 10 of 61
`
`after manufacturing) and paired with the Sphinx open-source speech recognition software. The
`
`Zentian accelerator demonstrated a two-times speedup of the Sphinx software.
`
`37.
`
`The implications of the Zentian team’s discovery were vast and addressed a number
`
`of shortcomings of speech recognition systems in the art that had stymied the technology’s
`
`commercial development. Speeding up the speech recognition process and reducing the overall
`
`power consumption would allow for the speech recognition system to run on a cheaper and/or
`
`lower-power processor and memory system. The Zentian accelerator would allow for the overall
`
`time the system might take to respond to spoken queries to be significantly reduced to a level more
`
`in-line with human conversations. Finally, Zentian’s accelerator approach would allow for the use
`
`of larger and more sophisticated acoustic models without increasing response time, facilitating
`
`better recognition accuracy and a larger recognizable vocabulary.
`
`38.
`
`Buoyed by their proof of concept, Catchpole and Larri attempted to market
`
`Zentian’s acoustic accelerator hardware. They spoke with representatives of leading speech
`
`recognition companies, such as Nuance, IBM, Sensory, Inc., and Samsung, as well as cellular
`
`carriers, such as Hutchison 3G UK Limited. Catchpole and Larri also attended the 2006 GSM
`
`World Congress in Barcelona, where they discussed their novel speech recognition SoC designs
`
`with an executive from Qualcomm. Unfortunately, the industry at the time remained focused on
`
`software solutions for smaller problems, such as voice-controlled phone dialing.
`
`39.
`
`In 2006, Zentian met with a director of Apple’s Silicon Engineering Group, the
`
`group responsible for the development of multiple classes of chips used in Apple mobile
`
`products.13 The presentation that Zentian shared with Apple identified problems with then-existent
`
`
`13 Apple describes its Silicon Technologies group as the team “responsible for the design and
`manufacture of our next generation of high-performance, power-efficient processor systems on
`
`
`-10-
`
`IPR2023-00034
`Apple EX1014 Page 10
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 11 of 61
`
`speech recognition systems, the promise of a dedicated hardware speech recognition accelerator,
`
`and the data demonstrating the success of Zentian’s own prototype. The presentation also
`
`identified that Zentian had filed patents in the United States (and in the United Kingdom) to cover
`
`its novel hardware-based approach to speech recognition systems.
`
`40.
`
`Apple listened to Zentian’s pitch and then ceased communication with the start-up.
`
`ZENTIAN’S PATENTS
`
`41.
`
`As a result of its work in the speech recognition field, Zentian has a patent portfolio
`
`that covers a range of inventions that improve the performance and efficiency of speech
`
`recognition on-device systems, including in the areas of parallel processing and circuit
`
`configurations. Five of Zentian’s patents, originating from two separate patent families, are at issue
`
`in this case.
`
`42.
`
`U.S. Patent No. 7,979,277 (the “’277 patent”) is entitled “Speech Recognition
`
`Circuit and Method” and issued on July 12, 2011. A true and correct copy of the ’277 patent is
`
`attached as Exhibit A to this Complaint.
`
`43.
`
`Under 35 U.S.C. § 154(b), the ’277 patent is entitled to 1,043 days of Patent Term
`
`Adjustment, extending its expiry date until July 2028.
`
`44.
`
`Zentian is the owner of all rights, title, and interest in and to the ’277 patent, with
`
`the full and exclusive right to file suit and enforce the ’277 patent, including the right to recover
`
`damages for past infringement.
`
`45.
`
`The ’277 patent is valid and enforceable under U.S. patent laws.
`
`
`chip, as well as the analog IP and high-speed interface evaluation those chips need.” See Careers
`at Apple: Hardware,
`Impossible
`is our
`favorite
`starting point, https://www.ap-
`ple.com/careers/us/hardware.html.
`
`-11-
`
`IPR2023-00034
`Apple EX1014 Page 11
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 12 of 61
`
`46.
`
`The ’277 patent claims are directed to a patent-eligible, non-abstract idea. They
`
`address, among other things, a specific improvement to the way in which conventional software-
`
`based speech recognition systems operated on electronic devices. The ’277 patent claims include
`
`an architecture for a speech recognition circuit in which data is “pipelined” between different
`
`hardware componentry. The claims specify materials (including types of processors, accelerator
`
`interfaces, etc.), structures, and configurations of the claimed speech recognition circuit designed
`
`to enhance power consumption and efficiency of the system and allow for real-time processing of
`
`more complex and robust language models.
`
`47.
`
`U.S. Patent No. 7,587,319 (the “’319 patent”) is entitled “Speech Recognition
`
`Circuit Using Parallel Processors” and issued on September 8, 2009. A true and correct copy of
`
`the ’319 patent is attached as Exhibit B to this Complaint.
`
`48.
`
`Under 35 U.S.C. § 154(b), the ’319 patent is entitled to 1,137 days of Patent Term
`
`Adjustment, extending its expiry date until March 2026.
`
`49.
`
`Zentian is the owner of all rights, title, and interest in and to the ’319 patent, with
`
`the full and exclusive right to file suit and enforce the ’319 patent, including the right to recover
`
`damages for past infringement.
`
`50.
`
`51.
`
`The ’319 patent is valid and enforceable under U.S. patent laws.
`
`The ’319 patent claims are directed to a patent-eligible, non-abstract idea. They
`
`address, among other things, a specific improvement to the way in which conventional software-
`
`based speech recognition systems operated on electronic devices. The ’319 patent claims include
`
`novel uses of a plurality of lexical tree processors connected in parallel to distribute the task of
`
`performing word recognition across processors.14 This architecture provides for improved and
`
`
`14 A “lexical tree” is a data structure that comprises a model of words.
`
`-12-
`
`IPR2023-00034
`Apple EX1014 Page 12
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 13 of 61
`
`faster processing, solving one of the problems of speech recognition known in the art—the
`
`computationally intensive search process that attempts to find the most likely set of words from a
`
`given lexicon. The claims specify materials (including types of processors, accelerator interfaces,
`
`etc.), structures, and configurations designed to enhance power consumption and efficiency of the
`
`system and allow for real-time processing of more complex and robust speech models.
`
`52.
`
`U.S. Patent No. 10,971,140 (the “’140 patent”) is entitled “Speech Recognition
`
`Circuit Using Parallel Processors” and issued on April 6, 2021. A true and correct copy of the ’140
`
`patent is attached as Exhibit C to this Complaint.
`
`53.
`
`Under 35 U.S.C. § 154(b), the ’140 patent is entitled to 0 days of Patent Term
`
`Adjustment, and its expiry date is in February 2023.
`
`54.
`
`Zentian is the owner of all rights, title, and interest in and to the ’140 patent, with
`
`the full and exclusive right to file suit and enforce the ’140 patent, including the right to recover
`
`damages for past infringement.
`
`55.
`
`56.
`
`The ’140 patent is valid and enforceable under U.S. patent laws.
`
`The ’140 patent claims are directed to a patent-eligible, non-abstract idea. They
`
`address, among other things, a specific improvement to the way in which conventional software-
`
`based speech recognition systems operated on electronic devices. The ’140 patent claims include
`
`a speech recognition circuit that has one or more clusters of processors, each of which comprises
`
`a plurality of processors and an acoustic model memory for storing acoustic model data. The
`
`disclosed and claimed circuit is further configured to generate an initial score for an audio sample,
`
`which is then used to determine whether to continue processing to determine a final score by
`
`processing a larger amount of model data than that which was processed to generate the initial
`
`score. The claims specify materials (including types of processors, accelerator interfaces, etc.),
`
`-13-
`
`IPR2023-00034
`Apple EX1014 Page 13
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 14 of 61
`
`structures, and configurations of the claimed speech recognition circuit designed to enhance power
`
`consumption and efficiency of the system and allow for real-time processing of more complex and
`
`robust language models.
`
`57.
`
`U.S. Patent No. 10,062,377 (the “’377 patent”) is entitled “Distributed Pipelined
`
`Parallel Speech Recognition System” and issued on August 28, 2018. A true and correct copy of
`
`the ’377 patent is attached as Exhibit D to this Complaint.
`
`58.
`
`Under 35 U.S.C. § 154(b), the ’377 patent is entitled to 0 days of Patent Term
`
`Adjustment, and its expiry date is in September 2025.
`
`59.
`
`Zentian is the owner of all rights, title, and interest in and to the ’377 patent, with
`
`the full and exclusive right to file suit and enforce the ’377 patent, including the right to recover
`
`for damages for past infringement.
`
`60.
`
`61.
`
`The ’377 patent is valid and enforceable under U.S. patent laws.
`
`The ’377 patent claims are directed to a patent-eligible, non-abstract idea. They
`
`address, among other things, a specific improvement to the way in which conventional software-
`
`based speech recognition systems operated on electronic devices. The ’377 patent claims include,
`
`among other things, a speech recognition system comprising a device programmed to calculate a
`
`feature vector from a digital audio stream (wherein the feature vector comprises a plurality of
`
`extracted and/or derived quantities from the stream during a defined audio time frame); a second
`
`device programmed to calculate distances indicating the similarity between the calculated feature
`
`vector and a plurality of acoustic states of an acoustic model; and a third device programmed to
`
`identify spoken words in the digital audio stream using Hidden Markov Models and/or Neural
`
`Networks. The system also includes a search stage for using the calculated distances to identify
`
`words within a lexical tree (the lexical tree comprising a model of words). The claims specify
`
`-14-
`
`IPR2023-00034
`Apple EX1014 Page 14
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 15 of 61
`
`materials (including types of processors, accelerator interfaces, etc.), structures, and configurations
`
`of the claimed speech recognition system designed to reduce power consumption and improve
`
`efficiency of the system and to allow for real time processing of more complex and robust language
`
`models.
`
`62.
`
`U.S. Patent No. 10,839,789 (the “’789 patent”) is entitled “Speech Recognition
`
`Circuit and Method” and issued on November 17, 2020. A true and correct copy of the ’789 patent
`
`is attached as Exhibit E to this Complaint.
`
`63.
`
`Under 35 U.S.C. § 154(b), the ’789 patent is entitled to 0 days of Patent Term
`
`Adjustment, and its expiry date is in September 2025.
`
`64.
`
`Zentian is the owner of all rights, title, and interest in and to the ’789 patent, with
`
`the full and exclusive right to file suit and enforce the ’789 patent, including the right to recover
`
`for past infringement.
`
`65.
`
`66.
`
`The ’789 patent is valid and enforceable under U.S. patent laws.
`
`The ’789 patent claims are directed to a patent-eligible, non-abstract idea. They
`
`address, among other things, a specific improvement to the way in which conventional software-
`
`based speech recognition systems operated on electronic devices. The ’789 patent claims include
`
`an acoustic coprocessor for processing data associated with an audio signal that includes an
`
`interface for receiving at least one feature vector, a calculating apparatus for calculating distances
`
`indicating a similarity between the feature vector and acoustic state of an acoustic model, and a
`
`second interface for sending at least one distance calculated by the calculating apparatus. The
`
`claims specify materials (including types of processors, accelerator interfaces, etc.), structures, and
`
`configurations of acoustic coprocessors designed to enhance power consumption and efficiency of
`
`the system and allow for real time processing of more complex and robust language models.
`
`-15-
`
`IPR2023-00034
`Apple EX1014 Page 15
`
`

`

`Case 6:22-cv-00122-ADA Document 1 Filed 02/02/22 Page 16 of 61
`
`67.
`
`For ease of reference, the ’277, ’377, and ’789 patents are part of the same patent
`
`family. The ’319 and ’140 patents are part of the same patent family.
`
`APPLE’S UNAUTHORIZED USE OF ZENTIAN’S TECHNOLOGY
`
`68.
`
`Apple’s infringing speech recognition solutions are central to shaping the user
`
`experience of many Apple products, including in Accused Products like Apple’s iPhone, iPad,
`
`Watch, HomePod smart speakers, Apple TV set top boxes, and Mac personal computers.
`
`69.
`
`First introduced in 2011, Siri uses speech recognition and natural language
`
`processing to enable Apple devices to play music, control smart home accessories, answer general
`
`knowledge inquiries, send messages, set clocks and timers, obtain news and weather information,
`
`and even get traffic reports and translations. According to Apple, “Siri lets you stay connected
`
`without lifting a finger” and is a “faster, easier way to do all kinds of useful things.”15
`
`70.
`
`Siri fulfills more than 25 billion requests each month, according to Apple’s director
`
`of Siri data science and engineering.16 Siri is, in turn, a core feature built into almost every device
`
`Apple sells, and critical to the unique user-device