PTAB

IPR2023-00119

Sonos Inc v. Google LLC

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Systems and Techniques for Speech Recognition
  • Brief Description: The ’330 patent discloses systems and techniques for managing multiple voice-controlled devices that may simultaneously detect the same hotword utterance from a user. The invention addresses the "duplicate response" problem by implementing a real-time coordination protocol, wherein devices exchange messages to collectively determine which single, most appropriate device should exit a low-power state and respond to the user's command.

3. Grounds for Unpatentability

Ground 1: Anticipation of Claims 1-3, 5, 7, 9-11, 13, 15, and 17-18 by Rosenberger

  • Prior Art Relied Upon: Rosenberger (Patent 8,340,975).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued Rosenberger anticipates every limitation of the challenged independent claims (1, 9, and 17) and their asserted dependents. Rosenberger is directed to the same technical problem as the ’330 patent: solving the "duplicate response" issue in a multi-device, voice-controlled environment. Petitioner asserted that Rosenberger’s system of "control devices" inherently meets the low power mode limitation, as it describes devices operating in a "low power ‘listening' mode" while awaiting a "hands-free trigger phrase" (the claimed hotword). Upon hearing the trigger phrase, each device calculates a "numeric weighted signal" (WS) that quantifies the recognition 'quality' (mapping to the claimed hotword score). The devices then broadcast their WS values as messages to each other while remaining in the low-power listening state. A single device determines it is best positioned to interact with the user by comparing its own WS message to the messages it receives from other devices. Only the device with the highest WS value exits the low-power mode to handle the voice command, thereby satisfying the limitation of determining to exit the low-power mode based on its own message and an additional message from another device. Dependent claims were also argued to be anticipated, such as by Rosenberger's disclosure of performing speech recognition after the winning device exits the low-power mode.

Ground 2: Obviousness of Claims 4, 6, 12, and 14 over Rosenberger in view of Basye

  • Prior Art Relied Upon: Rosenberger (Patent 8,340,975) and Basye (Application # 2014/0163978).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued this combination renders obvious the additional limitations in certain dependent claims. Specifically, for claims 4 and 12, which require determining a hotword without performing automated speech recognition processing, Basye explicitly teaches this concept for power conservation. Basye discloses a voice-processing chain where a low-power "keyword" detector (e.g., a "speech processing module") operates independently of a separate, more power-intensive "automatic speech recognition engine" that is only activated after a keyword is detected. For claims 6 and 14, which add transmitting a message based on a hotword score satisfying a threshold, Basye teaches calculating a "score or confidence level" for a detected keyword and comparing it to a predefined threshold. This determination is then used as a basis for activating other system components, such as a "network interface module" to transmit data. This directly teaches the claimed thresholding step.
    • Motivation to Combine: A POSITA would combine these references for clear reasons of technical improvement and efficiency. Both Rosenberger and Basye operate in the same field of voice-controllable computing devices and address intelligent "waking" of devices. Rosenberger discloses that its "control devices" may be battery-powered, making power efficiency a critical design concern. Basye provides explicit techniques to "improve the energy efficiency of the computing device," particularly those relying on battery power. A POSITA seeking to implement an efficient version of Rosenberger's system would look to teachings like Basye to defer power-intensive automated speech recognition. Furthermore, incorporating Basye's thresholding logic would predictably reduce "false positive" hotword detections in Rosenberger's system, thereby conserving power and network bandwidth by preventing devices from unnecessarily broadcasting WS messages.
    • Expectation of Success: Combining the references would have been straightforward for a POSITA. It involved integrating a known power-saving feature (Basye's two-stage detection and thresholding) into a known multi-device arbitration system (Rosenberger) to achieve the predictable benefits of enhanced battery life and system efficiency.

4. Arguments Regarding Discretionary Denial

  • Petitioner presented extensive arguments that discretionary denial would be inappropriate. Under 35 U.S.C. §314(a) and the Fintiv framework, Petitioner contended that the parallel district court litigation (the "NDCA Action") was stayed pursuant to 28 U.S.C. §1659, pending the final resolution of a related ITC proceeding. This stay meant that there had been minimal investment by the court and parties in the district court case, no trial date was set, and there was no risk of overlapping or conflicting decisions on validity. These factors, Petitioner argued, weigh strongly in favor of institution. Under 35 U.S.C. §325(d), Petitioner asserted that denial would be improper because the USPTO materially erred during prosecution by overlooking Rosenberger and Basye, which were not cited or considered by the Examiner. Petitioner argued these references are highly material and teach the core functionality of the challenged claims. The petition's reliance on a new expert declaration, evidence not previously available to the USPTO, was presented as another reason why the Board should review the patent's validity.

5. Relief Requested

  • Petitioner requests institution of an inter partes review and cancellation of claims 1-7, 9-15, and 17-18 of the ’330 patent as unpatentable.