DCT

8:18-cv-01293

SpeakWare Inc v. Microsoft Corp

I. Executive Summary and Procedural Information

  • Parties & Counsel:
  • Case Identification: 8:18-cv-01293, C.D. Cal., 07/26/2018
  • Venue Allegations: Plaintiff alleges venue is proper based on Defendant’s numerous regular and established places of business in the district, including Microsoft Store retail locations and corporate offices.
  • Core Dispute: Plaintiff alleges that Defendant’s voice-activated systems, including those incorporating the Cortana virtual assistant, infringe a patent related to hands-free, voice-operated remote control of electronic appliances.
  • Technical Context: The technology at issue falls within the domain of voice-activated remote controls, a foundational technology for the modern smart home and digital personal assistant market.
  • Key Procedural History: The complaint alleges that Defendant has been aware of the patent-in-suit since at least May 27, 2011, citing prosecution records of Microsoft’s own patents where the asserted patent was discussed. This alleged pre-suit knowledge forms the basis of the willfulness allegations.

Case Timeline

Date Event
1999-12-22 U.S. Patent No. 6,397,186 Priority Date
2002-05-28 U.S. Patent No. 6,397,186 Issues
2011-05-27 Alleged date of Microsoft's first knowledge of the patent
2018-07-26 Complaint Filed

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

U.S. Patent No. 6,397,186 - Hands-Free, Voice-Operated Remote Control Transmitter

  • Patent Identification: U.S. Patent No. 6,397,186, issued on May 28, 2002.

The Invention Explained

The patent addresses the shortcomings of prior art remote controls, which were typically hand-held and thus easily misplaced, inconvenient for users with physical limitations, and required manual interaction (’186 Patent, col. 1:24-35). Early voice-operated systems were not "truly hands-free" because they required the user to press a "talk switch" to activate listening, and some required physical connections to the controlled appliances (’186 Patent, col. 2:15-21, 2:42-49).

The invention is a wireless remote control system that can automatically switch from a low-power "sleep mode" to an "active mode" based on received voice commands, eliminating the need for a manual talk switch (’186 Patent, col. 7:12-16). This is achieved by using a sound activation circuit that detects an audio signal of sufficient amplitude to "wake up" the main speech recognition processor, which then processes the command and transmits a corresponding control signal to an appliance (’186 Patent, Abstract; col. 7:48-63).

This approach aimed to provide a "truly hands-free" remote operation that was more convenient and accessible while also conserving power by remaining in a low-power state until activated by voice (Compl. ¶ 27).

Key Claims at a Glance

The complaint asserts infringement of numerous claims, including independent claim 1 (Compl. ¶ 33).

The essential elements of independent claim 1 are:

  • An audio signal activated control system for controlling appliances.
  • A microphone for receiving and converting audio signals to electrical signals.
  • A speech recognition system with a processor, having two modes: a "low power sound activation mode" and a "speech recognition mode."
  • In the sound activation mode, the processor is in a "low power state."
  • The system is configured to "automatically switch" from the sound activation mode to the speech recognition mode "as a function of the amplitude of said electrical signals."
  • An appliance control circuit with a transmitter to receive control signals from the speech recognition system and transmit them to appliances.

III. The Accused Instrumentality

Product Identification

The accused instrumentalities are Microsoft’s voice-activated systems, including computing devices operating Microsoft Windows 10 and Harman Kardon Invoke smart speakers that utilize the Cortana virtual assistant, as well as associated Microsoft servers (Compl. ¶¶ 4, 22).

Functionality and Market Context

The accused systems use microphones to listen for voice commands, such as the "Hey Cortana" wake word (Compl. ¶ 11:25-26). The complaint alleges these systems operate in a low-power "sleep" state until a wake word is detected, at which point they switch to a full recognition mode to process user commands and control third-party appliances (Compl. ¶¶ 11:23-28, 12:1-15). A screenshot from Microsoft's website shows the Invoke speaker controlling smart home devices with the example command, "Hey Cortana, set the temperature to 70 degrees" (Compl. p. 14). The complaint alleges these systems control a range of appliances compatible with Cortana, as featured on Microsoft's website (Compl. ¶ 10:1-5).

IV. Analysis of Infringement Allegations

’186 Patent Infringement Allegations

Claim Element (from Independent Claim 1) Alleged Infringing Functionality Complaint Citation Patent Citation
An audio signal activated control system for controlling appliances comprising: The accused products are systems activated by audio signals (spoken words) for controlling appliances compatible with Cortana (Compl. ¶ 10:1-5). ¶33 col. 1:6-9
a microphone for receiving audio signals and converting said audio signals to electrical signals; The Harman Kardon Invoke smart speaker is alleged to include "seven far-field microphones." The complaint includes a diagram of the speaker identifying these microphones (Compl. p. 10). ¶33 col. 7:20-22
a speech recognition system for receiving said electrical signals, said speech recognition system including a processor and having a low power sound activation mode for detecting the presence of said electrical signals and a speech recognition mode for converting said electrical signals to electrical representative signals... wherein in said speech recognition mode said processor decodes said electrical representative signals and wherein in said sound activation mode said processor is in a low power state, The accused system allegedly has a "low power sound activation mode" (e.g., a "sleep" state) to detect wake words like "Hey Cortana," in which the processor consumes less power. It also has a "speech recognition mode" to recognize and process spoken commands after being woken up. ¶33 col. 7:12-16
said speech recognition system configured to automatically switch from said sound activation mode to said speech recognition mode as a function of the a-mplitude of said electrical signals The accused system is alleged to automatically switch from its low power mode to its speech recognition mode as a function of the amplitude of the electrical signals corresponding to the wake words "Hey Cortana." ¶33 col. 7:60-63
an appliance control circuit which includes a transmitter, said appliance control circuit configured to receive said control signals from said speech recognition system and generate and automatically transmit one or more appliance control signals to said one or more appliances. The accused system allegedly includes an "appliance control circuit" with a radio transceiver that receives control signals from the speech recognition components and transmits application control signals, such as Wi-Fi signals, to control appliances. ¶33 col. 8:10-17

Identified Points of Contention

  • Scope Questions: A primary question will be whether the claim term "as a function of the amplitude of said electrical signals" can be interpreted to cover the sophisticated phonetic analysis used in modern wake-word detection. The patent specification describes an embodiment that appears to rely on a simple signal amplitude threshold (’186 Patent, Fig. 2c), raising the question of whether Cortana's technology, which analyzes complex patterns beyond just loudness, falls within the claim's scope.
  • Technical Questions: The nature of the "low power state" will likely be contested. The defense may argue that the "always-on" listening architecture of a modern virtual assistant does not constitute the same "low power state" as the one described in the patent, which contemplates disabling the processor's clock source (’186 Patent, col. 12:18-24). The evidence will need to show a distinct operational state with verifiably lower power consumption as required by the claim.

V. Key Claim Terms for Construction

The Term: "automatically switch... as a function of the amplitude of said electrical signals"

  • Context and Importance: This term is the technological core of the infringement allegation. Its construction will determine whether the patent covers modern wake-word detection or is limited to simpler sound-level triggers. Practitioners may focus on this term because the alleged infringement depends on mapping Cortana's complex wake-word algorithm to this seemingly simple claim language.
  • Intrinsic Evidence for Interpretation:
    • Evidence for a Broader Interpretation: The specification states an "important aspect" is the ability to switch modes "solely by voice commands" to provide "true hands-free remote operation" (’186 Patent, col. 7:12-16). A party could argue that any system that uses characteristics of a voice command, including its amplitude profile, to switch modes falls within this purpose.
    • Evidence for a Narrower Interpretation: The detailed description of an embodiment explains that if an "amplified signal is of sufficient amplitude to activate the trigger circuit," the system wakes up (’186 Patent, col. 7:60-63). The associated Figure 2c depicts a relatively simple trigger circuit. A party could argue this language limits the claim to systems that trigger based on reaching a general volume threshold, not the specific phonetic pattern of a wake-word.

The Term: "low power state"

  • Context and Importance: This term is critical for both infringement and the patent's purported novelty in power saving. The dispute will likely center on what degree of power reduction or component deactivation is required to meet this limitation.
  • Intrinsic Evidence for Interpretation:
    • Evidence for a Broader Interpretation: The patent uses terms like "sleep mode" and "low power consumption mode" interchangeably with "sound activation mode" (’186 Patent, col. 7:13, 8:31-32). This may support an interpretation where any demonstrably reduced power consumption relative to the "speech recognition mode" is sufficient.
    • Evidence for a Narrower Interpretation: The specification describes placing the speech recognition circuit into a low power mode by "changing the logic states of said control register bits, removing the clock source 64 from the oscillator 63" (’186 Patent, col. 12:18-22). This could support a narrower construction requiring specific components, like the main oscillator, to be disabled.

VI. Other Allegations

  • Indirect Infringement: The complaint alleges induced infringement, stating that Microsoft encourages and instructs its customers, developers, and resellers to use the accused products in an infringing manner through materials like its website and product documentation (Compl. ¶¶ 35-37).
  • Willful Infringement: The complaint alleges willful infringement based on Microsoft's purported knowledge of the ’186 patent since at least May 27, 2011. This allegation is supported by claims that the patent was cited and discussed during the prosecution of Microsoft's own patent applications, and that Microsoft made responsive arguments regarding its disclosures (Compl. ¶¶ 40-41, 45).

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

  • A core issue will be one of technological scope: Can the claim limitation "as a function of the amplitude of said electrical signals," rooted in the patent's 1999 description of a sound-level trigger, be construed to read on the complex, phonetically-based wake-word detection algorithms used by modern virtual assistants like Cortana?
  • A second key issue will center on willfulness and pre-suit knowledge: The court will likely scrutinize the evidence presented from Microsoft's own patent prosecution files to determine whether Microsoft knew of the '186 patent and the relevance of its teachings to Microsoft's products, which will be dispositive for the claim of willful infringement and potential enhanced damages.