Amazon.com Inc v. Vocalife LLC

1. Case Identification

• Case #: IPR2021-01331
• Patent #: RE48,371
• Filed: July 30, 2021
• Petitioner(s): Amazon.com, Inc.
• Patent Owner(s): Vocalife LLC
• Challenged Claims: 22-41

2. Patent Overview

• Title: Microphone Array System for Enhancing a Target Sound Signal
• Brief Description: The ’371 patent relates to a system for enhancing a target sound signal, such as a human voice, using a microphone array and a digital signal processor (DSP). The DSP is described as implementing four functional units: a sound source localization (SSL) unit, an adaptive beamforming unit, a noise reduction unit, and an echo cancellation unit.

3. Grounds for Unpatentability

Ground 1: Obviousness over Reuss and Dmochowski - Claims 38-39 are obvious over Reuss in view of Dmochowski.

• Prior Art Relied Upon: Reuss (Patent 7,359,504) and Dmochowski (a 2007 IEEE article).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Reuss discloses a complete microphone array system with a DSP implementing three of the four claimed functional units: an adaptive beamforming unit, an echo cancellation unit ("echo controller"), and a noise reduction unit ("noise reducer"). Petitioner contended that Reuss’s adaptive beamformer, which dynamically steers its directional sensitivity toward a sound source, inherently performs sound source localization (SSL), thus teaching the fourth unit. Dmochowski was introduced to explicitly teach the specific method of "determining a delay" for the SSL unit, a limitation central to the challenged claims. Dmochowski provides a precise mathematical equation for calculating the time delay between microphones in an array as a function of the array's geometry and the sound's angle of arrival.
  • Motivation to Combine: A POSITA would combine Dmochowski with Reuss to implement a specific, effective SSL algorithm. Reuss describes the functional goal of tracking a sound source but does not disclose a particular algorithm for doing so. Dmochowski provides a well-known method (Steered Response Power, or SRP) and the necessary delay calculation to achieve this goal, making it a natural and logical addition to complete the system described in Reuss.
  • Expectation of Success: A POSITA would have had a high expectation of success. The SRP algorithm and delay calculations described by Dmochowski were well-understood and predictable. Reuss discloses using commercially available DSPs capable of implementing such algorithms, and combining the references would involve applying known engineering principles to achieve a predictable improvement in source tracking.

Ground 2: Obviousness over Reuss, Dmochowski, Li, and Brandstein - Claims 23-25, 27, 31-33, and 35 are obvious over the combination.

• Prior Art Relied Upon: Reuss (’504 patent), Dmochowski (2007 article), Li (a 2009 IEEE conference paper), and Brandstein (a 2001 textbook on microphone arrays).
• Core Argument for this Ground:
  • Prior Art Mapping: This ground builds upon the Reuss/Dmochowski combination by adding Brandstein to teach specific, advanced features of the adaptive beamforming unit recited in dependent claims. Petitioner argued that Brandstein, a seminal textbook, discloses the "most widely known" adaptive beamformer—the Griffiths-Jim beamformer (GJBF). The GJBF explicitly contains the claimed "fixed beamformer," "blocking matrix," and "adaptive filter." Brandstein also teaches using voice activity detection to control the adaptive filters, another claimed feature. For claims requiring the use of the SRP-PHAT algorithm, Brandstein was cited as teaching this as a more robust alternative to the standard SRP algorithm.
  • Motivation to Combine: A POSITA would combine Brandstein with the Reuss/Dmochowski system to implement a well-known, high-performance adaptive beamformer. Reuss itself suggests using "conventional or adaptive beamforming techniques well known to those of ordinary skill," directly inviting the inclusion of a prominent technique like the GJBF from Brandstein. Brandstein explains that the GJBF is effective at correcting steering vector errors, a known problem in systems like Reuss’s that rely on DOA estimates.
  • Expectation of Success: Success was expected because the GJBF was a widely known and predictable algorithm. Implementing this known component into Reuss’s system, as suggested by Reuss itself, would be a straightforward design choice for a POSITA seeking to improve performance.

Ground 3: Obviousness over Chen and Dmochowski - Claims 38-39 are obvious over Chen in view of Dmochowski.

• Prior Art Relied Upon: Chen (Application # 2006/0147063) and Dmochowski (2007 article).
• Core Argument for this Ground:
  • Prior Art Mapping: This ground presents Chen as an alternative primary reference to Reuss. Petitioner argued Chen discloses a complete system in a mobile phone that includes a microphone array and a DSP performing all four claimed functions. Specifically, Chen's DSP determines the DOA of a user's voice (SSL), uses adaptive beamforming to steer the array, includes an acoustic echo canceller (AEC) to reduce echo, and employs a noise suppressor. As in Ground 1, Dmochowski was combined to supply the explicit mathematical equation for calculating delays, which Chen discusses in principle but does not detail.
  • Motivation to Combine: A POSITA would combine Dmochowski's delay equation with Chen's system because Chen's disclosure presumes knowledge of how to calculate the DOA. Dmochowski provides the exact, well-known formula needed to implement this functionality, making it an obvious supplement to Chen's system description.
  • Expectation of Success: A POSITA would have expected success in this combination because it involves applying a standard, fundamental equation (from Dmochowski) to a system (Chen) that already relies on the principle behind that equation.
• Additional Grounds: Petitioner asserted additional obviousness challenges based on various combinations of Chen, Dmochowski, Li, Brandstein, and Abutalebi, relying on similar theories that a POSITA would combine known components for their stated advantages to achieve predictable results.

4. Key Claim Construction Positions

• Petitioner stated that for the purpose of the IPR proceeding, it adopted the claim constructions from a parallel district court litigation.
• "adaptive beamforming" was construed to mean "a beamforming process where the directivity pattern of the microphone array is capable of being adaptively steered in the direction of a target sound signal emitted by a target sound source in motion."
• "sound source localization unit" was given its plain and ordinary meaning, connoting "software/hardware in a DSP that includes functionality for locating a sound source."

5. Arguments Regarding Discretionary Denial

• Petitioner argued that discretionary denial under §314(a) based on the Fintiv factors would be inappropriate.
• Petitioner asserted that a stay of the parallel district court litigation is likely if the IPR is instituted, weighing against denial. To mitigate concerns of overlapping issues, Petitioner stipulated that it would not pursue in court any invalidity ground raised or reasonably available in this IPR.
• Petitioner further argued that the petition's merits are particularly strong because the primary references, Reuss and Chen, which disclose implementing all the claimed units in a single DSP, were never previously considered by the PTO, any court, or a jury.

6. Relief Requested

• Petitioner requests institution of an inter partes review and cancellation of claims 22-41 of Patent RE48,371 as unpatentable under 35 U.S.C. §103.