Dialect LLC v. Comcast Corportion

I. Executive Summary and Procedural Information

• Parties & Counsel:
  • Plaintiff: Dialect, LLC (Texas)
  • Defendant: Comcast Corporation (Pennsylvania); Comcast Cable Communications, LLC d/b/a Xfinity (Delaware); Comcast Cable Communications Management, LLC (Delaware); and Comcast of Northern Illinois, Inc. (Illinois)
  • Plaintiff's Counsel: Gerasimow Law; BLUE PEAK LAW GROUP LLP
• Case Identification: 1:25-cv-10341, N.D. Ill., 04/08/2026
• Venue Allegations: Plaintiff alleges venue is proper because Defendant maintains regular and established physical places of business in the Northern District of Illinois and has committed acts of patent infringement in the District.
• Core Dispute: Plaintiff alleges that Defendant's Xfinity Voice Remote infringes patents related to natural language understanding and voice recognition technology.
• Technical Context: The technology relates to systems that interpret natural human speech to execute commands and retrieve information, a foundational technology for modern voice assistants.
• Key Procedural History: The complaint alleges that the inventor of the asserted patents, VoiceBox Technologies, had a multi-year business relationship with Defendant, including presentations on its "patented voice interaction technology," a proposed license, and an alpha test of its voice remote technology with Comcast employees. Plaintiff also notes that a Delaware jury found Amazon's "Alexa" platform utilized VoiceBox's patented technology. Additionally, the complaint states that Google filed inter partes review (IPR) petitions against the '209, '039, '825, and '549 patents in April 2024, and that the Patent Trial and Appeal Board (PTAB) denied institution for the '209, '039, and '549 patents in October 2024.

Case Timeline

Date	Event
2002-06-03	Earliest Priority Date ('209, '039, '825, '549 Patents)
2005-08-29	Earliest Priority Date ('607 Patent)
2008-07-08	'209 Patent Issued
2009-11-17	'549 Patent Issued
2012-09-06	Meeting between VoiceBox and Comcast
2012-10-19	VoiceBox provided draft license to Comcast
2013-05-21	'607 Patent Issued
2015-01-01	Xfinity Voice Remote First Released (approx. date)
2016-02-16	'039 Patent Issued
2017-08-15	'825 Patent Issued
2024-04-01	Google filed IPR petitions ('209, '039, '825, '549 Patents) (approx. date)
2024-10-01	PTAB denied institution of IPRs ('209, '039, '549 Patents) (approx. date)
2026-04-08	Complaint Filed

II. Technology and Patent(s)-in-Suit Analysis

U.S. Patent No. 7,398,209 - "Systems And Methods For Responding To Natural Language Speech Utterance"

• Patent Identification: U.S. Patent No. 7,398,209, "Systems And Methods For Responding To Natural Language Speech Utterance," issued July 8, 2008.

The Invention Explained

• Problem Addressed: The patent's background describes the fundamental incompatibility between natural human speech, which relies heavily on context and domain knowledge, and machine-based queries, which are typically highly structured and not inherently natural ʼ209 Patent, col. 1:27-35
• The Patented Solution: The invention provides a method to process a natural language speech utterance by recognizing words, parsing them to determine a meaning and context, and then selecting an appropriate "domain agent" to handle the request ʼ209 Patent, abstract These domain agents are described as autonomous executables that organize domain-specific behavior and information, allowing the system to invoke the proper resources to process a user's request ʼ209 Patent, col. 2:48-54 Figure 6 illustrates this process flow from parsing the query to selecting an agent and formatting a command ʼ209 Patent, Fig. 6
• Technical Importance: The patent describes an architectural approach for making voice interaction systems more flexible and powerful by moving beyond rigid command-and-control systems to a more dynamic, context-aware framework using specialized software agents Compl. ¶22

Key Claims at a Glance

• The complaint asserts at least Claim 1 Compl. ¶71
• Claim 1 (Independent): A method responsive to a user's natural language speech utterance, comprising the essential elements of:
  • Receiving a user's natural language speech utterance containing a request.
  • Maintaining a dynamic set of prior probabilities.
  • Recognizing words and phrases using dictionary and phrase tables.
  • Parsing the words to determine a meaning and context for the request.
  • Selecting at least one "domain agent" based on the determined meaning, where the agent is an autonomous executable.
  • Formulating the request in accordance with a grammar used by the selected domain agent.
  • Invoking the domain agent to process the request.
  • Presenting the results of the processed request to the user.
• The complaint does not explicitly reserve the right to assert dependent claims for this patent.

U.S. Patent No. 8,447,607 - "Mobile Systems And Methods Of Supporting Natural Language Human-Machine Interactions"

• Patent Identification: U.S. Patent No. 8,447,607, "Mobile Systems And Methods Of Supporting Natural Language Human-Machine Interactions," issued May 21, 2013.

The Invention Explained

• Problem Addressed: The patent addresses the need to improve the reliability of voice recognition systems by better determining the context of a user's request, particularly in mobile environments that can accept both speech and non-speech inputs ʼ607 Patent, abstract '607 Patent, col. 21:49-60
• The Patented Solution: The patent describes a method for processing "multi-modal" inputs, which include both a natural language utterance and a non-speech input ʼ607 Patent, abstract The system generates transcriptions from both inputs, creates a user-specific "cognitive model" based on prior interactions, and merges the inputs to determine a request ʼ607 Patent, col. 15:18-40 This allows the system to, for example, display non-speech search results on a screen while concurrently accepting speech commands to execute requests on those results (ʼ607 Patent, col. 21:49-60).
• Technical Importance: This technology describes a way to create a more integrated user experience where voice and graphical interfaces work together, using personal history to improve accuracy Compl. ¶41

Key Claims at a Glance

• The complaint asserts at least Claim 12 Compl. ¶90
• Claim 12 (Independent): A method for processing natural language inputs, comprising the essential elements of:
  • Receiving a "multi-modal natural language input" from a user, which includes both a natural language utterance and a non-speech input.
  • Generating a non-speech transcription from the non-speech input.
  • Identifying the user.
  • Generating a speech-based transcription based on a "cognitive model" associated with the user, which includes information on prior interactions.
  • Generating a "merged transcription" from the speech and non-speech transcriptions.
  • Identifying an entry in a "context stack" that matches information in the merged transcription.
  • Identifying a "domain agent" associated with that context stack entry.
  • Determining a request based on the merged transcription and communicating it to the domain agent.
• The complaint does not explicitly reserve the right to assert dependent claims for this patent, but references "one or more claims" Compl. ¶90

Multi-Patent Capsules

• U.S. Patent No. 9,263,039

  • Patent Identification: U.S. Patent No. 9,263,039, "Systems And Methods For Responding To Natural Language Speech Utterance," issued February 16, 2016.
  • Technology Synopsis: The patent describes a method for processing both speech and non-speech communications by transcribing and merging them into a single query ʼ039 Patent, abstract The system then compares text combinations from the query to a "context description grammar," generates a relevance score, selects one or more domain agents based on that score, and arranges the resulting content to generate a response ʼ039 Patent, claim 13
  • Asserted Claims: At least Claim 13 Compl. ¶109
  • Accused Features: The complaint alleges that the Xfinity Voice Remote's functionality for processing voice commands infringes this patent Compl. ¶109

• U.S. Patent No. 9,734,825

  • Patent Identification: U.S. Patent No. 9,734,825, "Methods and Apparatus for Determining a Domain Based on the Content and Context of a Natural Language Utterance," issued August 15, 2017.
  • Technology Synopsis: The patent discloses a method for determining the correct domain for a user's speech utterance, even when the utterance is incomplete or ambiguous ʼ825 Patent, col. 1:32-40 The system uses a speech recognition engine, receives keywords and probabilities from domain agents, scores at least two possible contexts, and then determines a domain based on the recognized words and the context scores before selecting an agent to process the request ʼ825 Patent, abstract
  • Asserted Claims: At least Claim 5 Compl. ¶128
  • Accused Features: The complaint alleges that the Xfinity Voice Remote's functionality for processing voice commands infringes this patent Compl. ¶128

• U.S. Patent No. 7,620,549

  • Patent Identification: U.S. Patent No. 7,620,549, "System and method of supporting adaptive misrecognition in conversational speech," issued November 17, 2009.
  • Technology Synopsis: The patent describes a method for handling speech misrecognition by monitoring user actions after a voice command is processed ʼ549 Patent, abstract The system receives an initial voice utterance and then a subsequent "follow-up multimodal input." An "adaptive misrecognition engine" determines that the initial interpretation was incorrect if the follow-up input occurs "proximate in time" to the prior utterance, suggesting the user is taking corrective action ʼ549 Patent, claim 4
  • Asserted Claims: At least Claim 4 Compl. ¶147
  • Accused Features: The complaint alleges that the Xfinity Voice Remote's functionality for processing voice commands infringes this patent Compl. ¶147

III. The Accused Instrumentality

Product Identification

• The accused products are Defendant's Xfinity Voice Remote (models XR11, XR15, XR16, and XRA) and associated services and functionalities (the "Accused Product") Compl. ¶8 Compl. ¶66

Functionality and Market Context

• The Accused Product is a remote control that allows users to issue voice commands to control their cable service, such as changing channels, searching for shows, and viewing recommendations Compl. ¶67 The complaint alleges the remote "supports natural language speech recognition, so you can search by keyword or phrase" Compl. p. 26 The complaint includes a screenshot from Defendant's website illustrating example voice commands such as "Watch NBC," "Find kids movies," and "Show me what's on tonight at seven" Compl. p. 26
• The complaint alleges that Defendant first released the Xfinity Voice Remote in 2015 and had delivered over 18 million units by 2018 Compl. ¶66

IV. Analysis of Infringement Allegations

The complaint alleges direct infringement, induced infringement, and contributory infringement of the Asserted Patents Compl. ¶75 Compl. ¶94 Compl. ¶113 Compl. ¶132 Compl. ¶151 The complaint's specific infringement theories rely on appendices that were referenced but not provided with the complaint Compl. ¶74 Compl. ¶93 Compl. ¶112 Compl. ¶131 Compl. ¶150 Therefore, a detailed claim-chart style analysis is not possible based on the provided document.

The general infringement theory is that the system architecture and methods used by the Xfinity Voice Remote to receive, process, and respond to natural language voice commands from a user embody the methods claimed in the Asserted Patents. The complaint alleges that when a user speaks a command into the remote, the voice commands are sent for processing, which involves improving the voice recognition algorithms Compl. p. 26 This processing is alleged to practice the patented methods of using domain agents, cognitive models, context stacks, and multi-modal inputs to understand and respond to user requests Compl. ¶71 Compl. ¶90 Compl. ¶109 Compl. ¶128 Compl. ¶147 A screenshot in the complaint shows a desktop application from VoiceBox's early "Cybermind" technology, which allowed for both typed and spoken user input to retrieve information, demonstrating the multi-modal concepts central to several of the patents Compl. p. 8

• Identified Points of Contention:
  • Scope Questions: A potential area of dispute may be whether the server-side architecture that processes Comcast's voice commands, which is not fully detailed in the complaint, constitutes the claimed "domain agent," "cognitive model," and "context stack" as those terms are used in the patents.
  • Technical Questions: A key question will concern the specific algorithms and data structures used in the accused system. For example, regarding the '607 Patent, it may be disputed whether the accused system generates a "merged transcription" from distinct speech and non-speech inputs in the manner claimed. For the '549 Patent, a question may arise as to whether the accused system includes a specific "adaptive misrecognition engine" that determines an interpretation was incorrect based on the timing of a subsequent user action.

V. Key Claim Terms for Construction

For the '209 Patent:

• The Term: "domain agent"
• Context and Importance: This term is the core of Claim 1 of the '209 Patent. The definition of what constitutes a "domain agent"-specifically, an "autonomous executable that receives, processes, and responds to requests"-will be central to determining whether Comcast's backend software architecture infringes.
• Intrinsic Evidence for Interpretation:
  • Evidence for a Broader Interpretation: The specification describes agents as organizing "domain specific behavior and information" and as "complete, convenient and re-distributable packages or modules of functionality" ʼ209 Patent, col. 2:48-54 This language could support a broad definition covering various software modules that handle specific tasks (e.g., a "weather" module, a "movies" module).
  • Evidence for a Narrower Interpretation: The description that agents are "autonomous executables" and can be "distributed and redistributed" on various media or over networks could suggest a more constrained definition, potentially implying self-contained, independently deployable software packages rather than just components of a monolithic server application ʼ209 Patent, col. 2:50-51 '209 Patent, col. 2:62-67

For the '607 Patent:

• The Term: "cognitive model"
• Context and Importance: This term is critical to Claim 12 of the '607 Patent, which requires generating a transcription "based on a cognitive model associated with the user, wherein the cognitive model includes information on one or more prior interactions between the user and the device." The dispute will likely focus on whether the accused system's use of user history qualifies as the claimed "cognitive model."
• Intrinsic Evidence for Interpretation:
  • Evidence for a Broader Interpretation: The patent abstract states the invention "creates, stores, and uses extensive personal profile information for each user" to improve reliability ʼ607 Patent, abstract This suggests any system that stores and applies user-specific interaction history to aid recognition could fall within the term's scope.
  • Evidence for a Narrower Interpretation: The patent also links the cognitive model to a "conversational speech analyzer" and a "personalized cognitive model" that may be used to predict user behavior ʼ607 Patent, Fig. 8 This could support a narrower construction requiring a specific predictive or analytical function beyond simple storage and retrieval of user history.

VI. Other Allegations

• Indirect Infringement: The complaint alleges inducement based on Defendant's active advertising and providing "detailed instructions to users to encourage infringement" Compl. ¶79 Compl. ¶98 The complaint points to Defendant's websites that provide "Voice command tips for the Xfinity Voice Remote" as evidence of such instructions Compl. ¶67
• Willful Infringement: The complaint alleges willful infringement based on Defendant's alleged pre-suit knowledge of VoiceBox's patented technology, stemming from a business relationship that began as early as 2012 Compl. ¶82 This relationship allegedly included meetings, a live demo, an alpha test of VoiceBox's remote with Comcast employees, and VoiceBox's provision of a draft license to Comcast, which was not signed Compl. ¶¶59-62 The complaint includes a slide from a 2015 VoiceBox presentation to Comcast that explicitly mentions its "Patented Contextual Voice and NLU technology" Compl. p. 23

VII. Analyst's Conclusion: Key Questions for the Case

1. A central issue will be one of historical knowledge and intent: To what extent did Defendant's pre-existing business relationship with the patents' original inventor, VoiceBox, provide it with pre-suit knowledge of the patented technology, and does its subsequent development of the Xfinity Voice Remote support the allegation of willful infringement?
2. A key technical question will be one of architectural correspondence: Does the software architecture used by Comcast to process voice commands-which the complaint alleges involves "processing" by a "contracted service provider" Compl. p. 26-map onto the specific "domain agent" and "cognitive model" structures required by the asserted patent claims?
3. An important procedural question will be the impact of prior adjudications: How might the alleged PTAB decisions to deny institution of IPRs on several of the asserted patents, and the prior jury verdict against Amazon's Alexa, influence the trajectory and potential settlement dynamics of this case?