DCT

3:22-cv-06727

Jawbone Innovations LLC v. Amazon.com Inc

Log In

I. Executive Summary and Procedural Information

  • Parties & Counsel:
  • Case Identification: 2:21-cv-00435, E.D. Tex., 02/28/2022
  • Venue Allegations: Plaintiff alleges venue is proper in the Eastern District of Texas based on Defendant’s operation of regular and established places of business, including multiple fulfillment centers and "Amazon Hub Lockers," within the district.
  • Core Dispute: Plaintiff alleges that Defendant’s smart home devices, including the Echo, Echo Buds, and Fire TV product lines, infringe nine U.S. patents related to acoustic noise suppression and voice activity detection technology.
  • Technical Context: The technology at issue involves methods for improving audio clarity in electronic devices by using multiple microphones and sensors to distinguish a user's speech from ambient background noise.
  • Key Procedural History: The Amended Complaint was filed on February 28, 2022, following an Original Complaint filed on November 29, 2021. The complaint alleges that the patents originated with AliphCom, later Jawbone, Inc., which developed the technology for military and consumer headsets before its liquidation in 2017. Plaintiff alleges Amazon was aware of the patents at least as of 2017, when Jawbone's patent portfolio was reportedly marketed to potential buyers including Amazon.

Case Timeline

Date Event
2001-05-30 U.S. Patent No. 7,246,058 Priority Date
2002-01-01 AliphCom wins DARPA contract (Year only provided)
2002-03-27 U.S. Patent No. 8,467,543 Priority Date
2004-01-01 AliphCom launches "Jawbone" headset (Year only provided)
2007-06-13 Priority Date for multiple asserted patents (’091, ’072, ’213, ’611, ’080, ’357, ’691)
2007-07-17 U.S. Patent No. 7,246,058 Issues
2008-01-01 AliphCom launches Bluetooth "Jawbone" (Year only provided)
2011-01-01 AliphCom changes name to Jawbone, Inc. (Year only provided)
2011-09-13 U.S. Patent No. 8,019,091 Issues
2012-10-02 U.S. Patent No. 8,280,072 Issues
2012-11-17 U.S. Patent No. 8,321,213 Issues (as alleged in complaint)
2012-12-04 U.S. Patent No. 8,326,611 Issues
2013-06-18 U.S. Patent No. 8,467,543 Issues
2013-08-06 U.S. Patent No. 8,503,691 Issues
2017-01-01 Jawbone, Inc. forced into liquidation (Year only provided)
2017-01-01 Amazon allegedly became aware of Patents-in-Suit (Year only provided)
2020-09-15 U.S. Patent No. 10,779,080 Issues
2021-09-14 U.S. Patent No. 11,122,357 Issues
2021-11-29 Original Complaint Filed
2022-02-28 Amended Complaint Filed

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

U.S. Patent No. 8,019,091 - “Voice Activity Detector (VAD)-Based Multiple-Microphone Acoustic Noise Suppression”

The Invention Explained

  • Problem Addressed: The complaint alleges that conventional noise suppression techniques are limited in their ability to perfectly distinguish between periods of user speech and periods of only background noise (Compl. ¶¶ 33-34). This ambiguity can limit the effectiveness of noise removal algorithms.
  • The Patented Solution: The invention claims to solve this problem by using a non-acoustic physiological sensor, such as an accelerometer, to achieve "perfect voice activity detection" that can sense vibration in human tissue associated with speech (Compl. ¶¶ 33-34). This VAD signal informs a noise removal algorithm which of two different "transfer functions" to use: one calculated when only noise is present, and another when speech is present, thereby improving noise suppression (Compl. ¶ 34; ’091 Patent, Abstract).
  • Technical Importance: This approach, which fuses data from acoustic microphones and a physiological sensor, claims to "effectively remove noise from a signal no matter how many noise sources are present" (Compl. ¶ 34).

Key Claims at a Glance

  • The complaint asserts at least independent claim 11 (Compl. ¶ 51).
  • The essential elements of Claim 11 are:
    • A system for removing acoustic noise comprising a receiver, at least one sensor, and a processor.
    • The receiver receives at least two acoustic signals from at least two microphones.
    • The sensor receives human tissue vibration information associated with voicing activity.
    • The processor generates a plurality of transfer functions, including a first transfer function and a second transfer function.
    • The first transfer function is representative of a ratio of energy of acoustic signals and is generated when voicing activity is absent.
    • The second transfer function is representative of the acoustic signals and is generated when voicing activity is present.
    • Acoustic noise is removed using the first transfer function and at least one combination of the first and second transfer functions to produce a denoised data stream.
  • The complaint does not explicitly reserve the right to assert dependent claims for this patent.

U.S. Patent No. 7,246,058 - “Detecting Voiced and Unvoiced Speech Using Both Acoustic and Nonacoustic Sensors”

The Invention Explained

  • Problem Addressed: The technical challenge is accurately distinguishing not just between speech and noise, but between different types of speech (voiced and unvoiced) and background noise (Compl. ¶ 67). Voiced speech (e.g., vowels) involves vocal cord vibration, while unvoiced speech (e.g., an "s" sound) does not, making them difficult to separate from noise using traditional methods.
  • The Patented Solution: The invention employs a multi-modal sensing system. It uses a "voicing sensor" to receive physiological information (e.g., tissue vibration) and generates "cross correlation data" between this physiological signal and an acoustic signal to identify voiced speech. Concurrently, it uses two acoustic microphones to generate "difference parameters" (representing relative signal gain) to distinguish between unvoiced speech and ambient noise based on different gain thresholds (’058 Patent, Abstract; Compl. ¶ 67).
  • Technical Importance: The technology provides a system for more granular and accurate classification of acoustic signals into voiced speech, unvoiced speech, or noise by combining physiological and multi-microphone acoustic data streams (Compl. ¶ 67).

Key Claims at a Glance

  • The complaint asserts at least independent claim 1 (Compl. ¶ 67).
  • The essential elements of Claim 1 are:
    • A system for detecting voiced and unvoiced speech comprising at least two microphones, at least one voicing sensor, and at least one processor.
    • The processor generates cross correlation data between the physiological information (from the voicing sensor) and an acoustic signal from one microphone.
    • The processor identifies information as voiced speech when the cross correlation data exceeds a correlation threshold.
    • The processor generates difference parameters between the acoustic signals from the two microphones, representing relative signal gain differences.
    • The processor identifies information as unvoiced speech when the difference parameters exceed a gain threshold.
    • The processor identifies information as noise when the difference parameters are less than the gain threshold.
  • The complaint does not explicitly reserve the right to assert dependent claims for this patent.

Multi-Patent Capsule: U.S. Patent No. 8,280,072

  • Patent Identification: U.S. Patent No. 8,280,072, “Microphone Array with Rear Venting,” issued October 2, 2012 (Compl. ¶ 19).
  • Technology Synopsis: The patent describes acoustic noise suppression using an array of physical microphones to form a smaller array of "virtual" microphones. The outputs of these virtual microphones are then combined to produce a signal with reduced acoustic noise (Compl. ¶ 36).
  • Asserted Claims: At least Claim 1 (Compl. ¶ 84).
  • Accused Features: The complaint alleges that Amazon products like the Amazon Echo, which comprises an array of at least seven physical microphones that form six beamformed virtual microphones, practice the claimed method (Compl. ¶¶ 37, 85-86). The complaint includes a teardown photograph of an Amazon Echo circuit board highlighting the location of seven microphones (Compl. p. 30).

Multi-Patent Capsule: U.S. Patent Nos. 8,321,213 and 8,326,611

  • Patent Identification: U.S. Patent No. 8,321,213 and U.S. Patent No. 8,326,611, both titled “Acoustic Voice Activity Detection (AVAD) for Electronic Systems,” issued November 17, 2012 and December 4, 2012, respectively (Compl. ¶¶ 20-21).
  • Technology Synopsis: These patents describe detecting voice activity by forming virtual microphones from an array of physical microphones. A ratio of energies between these virtual microphones is then compared to a threshold to determine when voicing is occurring (Compl. ¶ 38).
  • Asserted Claims: At least Claim 1 of the ’213 Patent and Claim 1 of the ’611 Patent (Compl. ¶¶ 96, 109).
  • Accused Features: The complaint alleges that Amazon Echo smart speakers infringe by using beamforming to create virtual microphones and detecting user speech (e.g., an "Alexa" wake word) by comparing a ratio of energies from these microphones to a threshold (Compl. ¶ 39). A presentation slide is provided showing a waveform of the "Alexa" wake word being distinguished from ambient noise (Compl. p. 13).

Multi-Patent Capsule: U.S. Patent No. 10,779,080

  • Patent Identification: U.S. Patent No. 10,779,080, “Dual Omnidirectional Microphone Array,” issued September 15, 2020 (Compl. ¶ 22).
  • Technology Synopsis: The patent describes noise suppression using an array of omnidirectional microphones to form virtual microphones engineered to have a similar response to noise but a dissimilar response to speech. This differential response allows an adaptive filter to reduce noise without significantly distorting the desired speech (Compl. ¶ 40).
  • Asserted Claims: At least Claim 1 (Compl. ¶ 123).
  • Accused Features: The complaint accuses Amazon products, including Echo Buds and Echo smart speakers, which allegedly comprise at least two physical omnidirectional microphones and a processing unit that generates at least two beamformed virtual microphones with the claimed response characteristics (Compl. ¶ 41).

Multi-Patent Capsule: U.S. Patent Nos. 11,122,357 and 8,503,691

  • Patent Identification: U.S. Patent No. 11,122,357, “Forming Virtual Microphone Arrays Using Dual Omnidirectional Microphone Array (DOMA),” issued September 14, 2021, and U.S. Patent No. 8,503,691, “Virtual Microphone Arrays Using Dual Omnidirectional Microphone Array (DOMA),” issued August 6, 2013 (Compl. ¶¶ 23, 25).
  • Technology Synopsis: These patents describe acoustic noise suppression where signals from physical or virtual microphones are combined by filtering and summing them in the time domain. This process applies a varying linear transfer function to suppress noise in the output signal (Compl. ¶ 42).
  • Asserted Claims: At least Claim 1 of the ’357 Patent and Claim 23 of the ’691 Patent (Compl. ¶¶ 138, 168).
  • Accused Features: The complaint alleges that Amazon Echo Buds and smart speakers infringe by comprising arrays of physical microphones whose outputs are combined into beamformed microphones to reduce signal noise (Compl. ¶ 43).

Multi-Patent Capsule: U.S. Patent No. 8,467,543

  • Patent Identification: U.S. Patent No. 8,467,543, “Microphone and Voice Activity Detection (VAD) Configurations for Use with Communications Systems,” issued June 18, 2013 (Compl. ¶ 24).
  • Technology Synopsis: The patent describes a system with a voice detection subsystem and a denoising subsystem. It specifies a microphone configuration where one microphone is oriented toward a talker's mouth and a second is oriented away, allowing the denoising system to subtract noise captured by the second microphone from the signal captured by the first (Compl. ¶ 44).
  • Asserted Claims: At least Claim 1 (Compl. ¶ 151).
  • Accused Features: The complaint accuses the Amazon Echo Buds, which allegedly have at least one microphone on each earbud oriented towards the user's mouth and another oriented away, and use accelerometers to detect tissue vibration for voice activity detection (Compl. ¶ 45).

III. The Accused Instrumentality

Product Identification

  • The "Accused Products" are identified as all versions and variants of Amazon Echo smart speakers, Echo Show smart displays, Fire TV Cube, and Echo buds (Compl. ¶ 46).

Functionality and Market Context

  • The complaint alleges these products incorporate noise suppression and voice activity detection technologies to enable features like wake-word detection for the Alexa virtual assistant (Compl. ¶ 39). The Amazon Echo Buds are specifically alleged to include "2 external beamforming microphones" and an accelerometer in each earbud to detect human tissue vibration (Compl. ¶ 35). The Amazon Echo smart speakers are alleged to use an array of at least seven physical microphones to create virtual beamformed microphones for isolating a user's voice from background noise (Compl. ¶ 37). The complaint references a presentation slide illustrating how beamforming is used to isolate the "Alexa" wake word from ambient noise (Compl. p. 13). Defendant is characterized as a "leading manufacturing and seller of virtual assistant smart speakers" (Compl. ¶ 4).

IV. Analysis of Infringement Allegations

'091 Patent Infringement Allegations

Claim Element (from Independent Claim 11) Alleged Infringing Functionality Complaint Citation Patent Citation
a receiver that receives at least two acoustic signals via at least two acoustic microphones positioned in a plurality of locations; The Echo Buds comprise a receiver that receives signals via a microphone array with at least two microphones. ¶53 col. 20:12-16
at least one sensor that receives human tissue vibration information associated with human voicing activity of a user; The Echo Buds comprise an accelerometer that receives human tissue vibration associated with voicing activity. ¶54 col. 20:17-19
a processor coupled among the receiver and the at least one sensor that generates a plurality of transfer functions, wherein the plurality of transfer functions includes a first transfer function representative of a ratio of energy of acoustic signals received using at least two different acoustic microphones... The Echo Buds comprise a processor (e.g., Realtek SoC, NXP DSP) that utilizes the microphone array to detect speech and generate a plurality of transfer functions, including a first transfer function based on energy ratio. ¶55 col. 20:20-29
...wherein the first transfer function is generated in response to a determination that voicing activity is absent from the acoustic signals for a period of time, The Echo Buds allegedly generate the first transfer function when the voice detecting accelerometer indicates that voicing activity is absent. ¶56 col. 20:29-32
...wherein the plurality of transfer functions includes a second transfer function representative of the acoustic signals, wherein the second transfer function is generated in response to a determination that voicing activity is present... The Echo Buds allegedly generate a second transfer function in response to the accelerometer detecting that voicing activity is present. ¶57 col. 21:1-5
...wherein acoustic noise is removed from the acoustic signals using the first transfer function and at least one combination of the first transfer function and the second transfer function to produce the denoised acoustic data stream. The Echo Buds allegedly remove noise by applying the first transfer function (generated when voicing is absent) and a combination of the first and second transfer functions (when voicing is detected). ¶58 col. 21:6-11
  • Identified Points of Contention:
    • Technical Questions: A central question may be whether the accused products' processor actually generates two distinct and separate "transfer functions" that are selected based on the binary state (present/absent) of the VAD sensor. An alternative technical approach could involve a single, continuously adaptive algorithm where the VAD signal is merely one of several weighted inputs, which may not meet the claim's requirement of generating discrete first and second functions in response to the VAD determination.
    • Scope Questions: The definition of "generates" will be critical. Does the processor's standard operation inherently "generate" these functions as claimed, or does the claim require a specific, discrete calculation step that is triggered by the VAD's output?

'058 Patent Infringement Allegations

Claim Element (from Independent Claim 1) Alleged Infringing Functionality Complaint Citation Patent Citation
at least two microphones that receive the acoustic signals; Each earbud of the Echo Buds comprises at least two outward-facing MEMS microphones. ¶68 col. 11:13-14
at least one voicing sensor that receives physiological information associated with human voicing activity; The Echo Buds comprise an accelerometer that receives human tissue vibration associated with voicing activity. ¶69 col. 11:15-17
at least one processor coupled among the microphones and the voicing sensor, wherein the at least one processor; The Echo Buds comprise a processor (e.g., Realtek SoC, NXP DSP) coupled between the microphones and accelerometer. ¶70 col. 11:18-20
generates cross correlation data between the physiological information and an acoustic signal received at one of the two microphones; The processor allegedly generates cross correlation data between tissue vibration and an acoustic signal from one microphone. ¶71 col. 11:21-24
identifies information of the acoustic signals as voiced speech when the cross correlation data...exceeds a correlation threshold; The processor allegedly identifies acoustic signals as speech when the cross correlation data exceeds a threshold based on vibration and/or acoustic signals. ¶72 col. 11:24-28
generates difference parameters between the acoustic signals received at each of the two receivers...representative of the relative difference in signal gain...; The processor allegedly generates difference parameters between signals from each microphone representing relative signal gain differences. ¶73 col. 11:29-34
identifies information of the acoustic signals as unvoiced speech when the difference parameters exceed a gain threshold; The processor allegedly identifies information as unvoiced speech when the difference parameter exceeds a gain threshold. ¶74 col. 11:34-36
and identifies information of the acoustic signals as noise when the difference parameters are less than the gain threshold. The processor allegedly identifies acoustic signals as noise when the difference parameters are less than the gain threshold. ¶75 col. 11:37-39
  • Identified Points of Contention:
    • Technical Questions: The infringement theory hinges on whether the accused processor executes the specific three-part logic of Claim 1: (1) using cross-correlation for voiced speech, (2) using difference parameters for unvoiced speech, and (3) using difference parameters for noise. It is possible the accused products use a more integrated method, such as a machine learning model, that does not perform these discrete, separable classification steps. The complaint's technical details image for the Echo Buds shows multiple microphones but does not detail the specific algorithm used (Compl. p. 9).
    • Scope Questions: The case may turn on the construction of "generates cross correlation data" and "generates difference parameters." Does this require the explicit performance of those specific mathematical operations as understood at the time of the invention, or could it be interpreted more broadly to cover any process that achieves the same classification result?

V. Key Claim Terms for Construction

  • The Term: "plurality of transfer functions" (’091 Patent, Claim 11)

  • Context and Importance: Claim 11 requires the processor to generate a plurality of functions: a "first transfer function" when voicing is absent and a "second transfer function" when voicing is present. The construction of this term is critical because modern adaptive systems may not generate discrete, separate functions for different states. Practitioners may focus on whether a single, continuously adapting algorithm that uses the VAD signal as an input can be said to generate a "plurality" of functions, or if the claim requires distinct operational modes.

  • Intrinsic Evidence for Interpretation:

    • Evidence for a Broader Interpretation: The specification states that the system "generates a transfer function representative of the received acoustic signals upon determining that voicing information is absent" and then removes noise using that function, suggesting a process tied to a specific state (’091 Patent, col. 16:11-20). This could be argued to describe the state of any adaptive filter at a given moment.
    • Evidence for a Narrower Interpretation: The claim's structure, which separately introduces the "first transfer function" (generated when voicing is absent) and the "second transfer function" (generated when voicing is present), suggests they are distinct entities created in response to different conditions. The abstract also describes generating a transfer function when voicing is absent and using it to remove noise, reinforcing the idea of a discrete function for a specific state (’091 Patent, Abstract).

  • The Term: "generates cross correlation data" (’058 Patent, Claim 1)

  • Context and Importance: The claim requires the processor to perform a specific mathematical step—generating cross correlation data—to identify voiced speech. The case may depend on whether Amazon's processor literally performs this calculation. Practitioners may focus on this term because if the accused products use a different technique (e.g., a neural network) to correlate the sensor and audio inputs, infringement could be avoided if this term is construed narrowly to its mathematical definition.

  • Intrinsic Evidence for Interpretation:

    • Evidence for a Broader Interpretation: The specification describes the system's goal as determining when "sensor data that correlates highly with the acoustic signal is declared as speech" (’058 Patent, col. 5:29-31). This focus on the functional goal of correlation could support an interpretation that covers any method achieving that result.
    • Evidence for a Narrower Interpretation: The claim language is specific, reciting the generation of "cross correlation data" which has a well-understood mathematical meaning. The detailed description refers to "the cross-correlation (XCORR) between the acoustic signal from microphone 1 and the sensor data," further tying the claim to this specific technical operation (’058 Patent, col. 5:2-4).

VI. Other Allegations

  • Indirect Infringement: The complaint alleges that Amazon induces infringement of all nine patents. The basis for this allegation is that Amazon provides "instruction manuals, websites, promotional materials, advertisements, and other information" that instruct customers and end-users how to use the Accused Products in a manner that directly infringes the patent claims (Compl. ¶¶ 60, 77, 89, 102, 116, 131, 144, 160, 176).
  • Willful Infringement: Willfulness is alleged for the ’543 and ’691 patents. The complaint alleges that Amazon has known of Jawbone's patents, including the ’543 patent, "at least since they were marketed to Amazon following Jawbone Inc.'s liquidation" (Compl. ¶ 162). For the ’691 patent, the complaint alleges Amazon has been aware of its infringement "since around the time it first became aware of the ’691 Patent" and has been "willfully blind" (Compl. ¶ 177).

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

  • A core issue will be one of algorithmic equivalence: do Amazon's accused audio processing systems, which may rely on complex, unified machine-learning models, perform the discrete, multi-step logical processes required by the claims? For example, does the system generate distinct "first" and "second" transfer functions based on a binary VAD state as claimed in the ’091 patent, or does it use a single, continuously adapting algorithm?
  • A central question will be one of definitional scope: can claim terms rooted in specific mathematical operations of the early 2000s, such as "generates cross correlation data," be construed to cover modern AI-based systems that may achieve a similar functional outcome without explicitly performing that named calculation? The construction of these terms will likely be dispositive for several of the asserted patents.
  • A key factual dispute will concern pre-suit knowledge: what evidence can Plaintiff produce to support its allegation that Amazon was aware of the asserted patents and their relevance to its products following the 2017 liquidation of Jawbone, Inc.? The outcome of this discovery will be critical to the claims for willful infringement.