Google LLC v. Buffalo Patents LLC

1. Case Identification

• Case #: IPR2023-01387
• Patent #: 8,204,737
• Filed: September 12, 2023
• Petitioner(s): Google LLC
• Patent Owner(s): Buffalo Patents LLC
• Challenged Claims: 1-16 and 19

2. Patent Overview

• Title: System and Method for Generating Text
• Brief Description: The ’737 patent discloses a system for generating text responsive to both voice and handwriting inputs. The system uses a plurality of user-specific message models and a local message model to recognize and convert user inputs into text data.

3. Grounds for Unpatentability

Ground 1: Obviousness of Claims 1, 4, 12-13, 15-16, and 19 over Ditzik, Tchorzewski, and Roberts

• Prior Art Relied Upon: Ditzik (Patent 6,167,376), Tchorzewski (Canadian Patent 1,335,002), and Roberts (Patent 5,027,406).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner asserted that Ditzik disclosed the foundational system for generating text from voice and handwriting inputs, including a microphone, a stylus on a tablet surface, and the use of acoustic, language, and handwriting models. However, Ditzik’s system was not inherently designed for multiple distinct users. Tchorzewski was argued to remedy this by teaching a multi-user speech recognition system that stores a plurality of user-specific "templates" (i.e., user message models) in a central database and selects the appropriate one upon activation of a user terminal. Roberts was cited for its disclosure of adapting and updating recognition models based on user input, such as correcting misrecognized words, which teaches the "adjusting" limitation of the claims.
  • Motivation to Combine: A POSITA would combine Ditzik’s multimodal input system with Tchorzewski’s multi-user architecture to improve its functionality and efficiency for different users of the same device. A POSITA would further incorporate Roberts’s method for updating language and acoustic models to increase the recognition accuracy for each specific user over time, a known goal in the field.
  • Expectation of Success: The combination involved applying known techniques (multi-user support, model adaptation) to a known system type to achieve the predictable result of a more accurate, personalized, and efficient recognition system.

Ground 2: Obviousness of Claims 1-2, 4, 12-13, and 16-19 over Root, Roberts, and Lewis

• Prior Art Relied Upon: Root (Patent 5,600,781), Roberts (Patent 5,027,406), and Lewis (Application # 2001/0053978).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner presented Root as disclosing a personalized computer interface system that uses a portable "personality profile" for each user, stored on a PCMCIA card, containing user-specific handwriting and voice recognition data. This taught the concept of selecting a particular user message model from a plurality of models corresponding to different users. Roberts, as in Ground 1, was argued to teach the adaptation of language and acoustic models based on user feedback. Lewis was cited as providing context, explaining that conventional recognition systems decode input using a plurality of trained, user-specific models.
  • Motivation to Combine: A POSITA would be motivated to enhance Root's system, which already stored user-specific models, by incorporating the dynamic model adjustment taught by Roberts. This would allow Root’s static "personality profiles" to be continuously trained and refined, improving overall recognition accuracy. Lewis confirmed that using such trained models was a standard approach.
  • Expectation of Success: Combining Root's user-profile system with Roberts's well-understood model training techniques was presented as a straightforward integration of known elements to achieve the predictable outcome of a more accurate, adaptive recognition system.

Ground 3: Obviousness of Claims 2-3, 5-6, 11, and 14 over the combination of Ditzik, Tchorzewski, and Roberts in view of Verrier

• Prior Art Relied Upon: Ditzik (Patent 6,167,376), Tchorzewski (Canadian Patent 1,335,002), Roberts (Patent 5,027,406), and Verrier (European Application # 0 622 724 A2).

• Core Argument for this Ground:

  • Prior Art Mapping: This ground built upon the base combination of Ditzik, Tchorzewski, and Roberts from Ground 1. Petitioner argued that Verrier taught specific hardware features for a stylus that were missing from Ditzik. Verrier disclosed a stylus containing an integrated microphone (mapping to claim 3), a proximity sensor to detect the stylus's closeness to a tablet (mapping to claim 5), and an identification circuit to validate the user of the stylus (mapping to claims 2 and 6). Verrier also taught switching between audio input and freehand input based on the stylus's movement toward the tablet (mapping to claim 14).
  • Motivation to Combine: A POSITA would improve the base system of Ditzik by substituting its basic stylus with the advanced, feature-rich stylus of Verrier. This modification would improve usability by housing the microphone in the stylus for better audio capture, automating the switch between input modes, and enabling user identification directly from the stylus to trigger the selection of the correct user model.
  • Expectation of Success: This combination involved the simple substitution of one known element (a basic stylus) for another (an advanced stylus) to gain known benefits, representing a predictable design improvement.

• Additional Grounds: Petitioner asserted additional obviousness challenges based on adding Bellegarda (for its teachings on syntactic and semantic language models) and Campbell (for its teachings on automatic speaker recognition for user identification) to the primary prior art combinations.

4. Arguments Regarding Discretionary Denial

• §314(a) (Fintiv): Petitioner argued that discretionary denial under Fintiv would be improper. It asserted that the related district court litigation against a third party (Motorola Mobility) was stayed, discovery had not commenced, and no trial date was set. Given these early stages and the strength of the petition's arguments, Petitioner contended that an IPR would be a more efficient resolution.
• §325(d): Petitioner argued that denial under §325(d) was inappropriate because the core prior art references forming its grounds (e.g., Roberts, Tchorzewski, Root, Lewis, Bellegarda, Campbell) were never considered by the Examiner during prosecution. For the art that was before the Examiner (Ditzik and Verrier), Petitioner argued the Examiner did not apply them in the combinations asserted in the petition and overlooked their most relevant teachings, constituting a material error.

5. Relief Requested

• Petitioner requested the institution of an inter partes review and the cancellation of claims 1-16 and 19 of Patent 8,204,737 as unpatentable under 35 U.S.C. §103.