UNITED STATES PATENT AND TRADEMARK OFFICE
`
`
`
`
`
`
`
`
`
`
`
`
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`
`
`
`
`
`
`
`
`
`
`
`
`APPLE INC.,
`Petitioner,
`
`v.
`
`ANDREA ELECTRONICS INC.,
`Patent Owner.
`
`Patent No. 6,363,345
`
`
`IPR2017-00626
`
`
`
`
`
`
`
`
`
`
`
`
`
`REPLY DECLARATION OF BERTRAND HOCHWALD
`REGARDING U.S. PATENT NO. 6,363,345
`
`
`
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. Cover
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`
`
`
`
`
`
`
`
`
`
`
`
`APPLE INC.,
`Petitioner,
`
`v.
`
`ANDREA ELECTRONICS INC.,
`Patent Owner.
`
`Patent No. 6,363,345
`
`
`IPR2017-00626
`
`
`
`
`
`
`
`
`
`
`
`
`
`REPLY DECLARATION OF BERTRAND HOCHWALD
`REGARDING U.S. PATENT NO. 6,363,345
`
`
`
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. Cover
`
`
`
`TABLE OF CONTENTS
`
`INTRODUCTION ......................................................................................... 1
`I.
`II. RESPONSE TO DR. DOUGLAS’S OPINIONS ........................................ 1
`A. Martin’s Sub-Windows Are an Optional Feature ................................. 1
`B. Martin Discloses the Claimed “Future Minimum” Even Where
`Sub-Windows Are Used ........................................................................ 5
`C.
`Well His Noise Floor Algorithm Works ............................................... 9
`D. Dr. Douglas Mischaracterizes My Deposition Testimony .................. 11
`E.
`Is a Long Time .................................................................................... 12
`
`Dr. Douglas Is Incorrect that Hirsch Teaches That 0.2 Seconds
`
`Dr. Douglas Misinterprets Martin’s Own Description of How
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. i
`
`
`
`I.
`
`INTRODUCTION
`I have been retained by counsel for Apple Inc. as an expert witness in
`1.
`
`the above-captioned proceeding. I have been asked to provide an opinion
`
`regarding the patentability of certain claims in U.S. Patent No. 6,363,345
`
`(“the ’345 Patent”) (Exhibit 1001). I previously submitted declarations in this
`
`matter as Exhibits 1003 and 1004.
`
`2. My background and qualifications are set forth in my opening
`
`declarations Exhibits 1003 and 1004. A copy of my CV was attached as Exhibit B
`
`to those declarations. I disclosed the compensation I am receiving, and prior
`
`testimony in my opening declarations. I also set forth my understanding of the
`
`relevant legal standards in my opening declarations.
`
`3.
`
`I understand that Andrea submitted responses to the petitions and the
`
`Institution decisions, and that Andrea submitted two declarations from Dr. Scott
`
`Douglas (both labeled Ex. 2002). I have considered Andrea’s Responses and Dr.
`
`Douglas’s declarations, and this declaration sets forth my reply to certain of
`
`Andrea’s and Dr. Douglas’s arguments.
`
`II. RESPONSE TO DR. DOUGLAS’S OPINIONS
`A. Martin’s Sub-Windows Are an Optional Feature
`Andrea and Dr. Douglas contend that multiple sub-windows (W > 1)
`4.
`
`are a “crucial” part of Martin’s algorithm. [Ex. 2002 (-626 Douglas) at ¶60.] They
`
`are wrong because sub-window(s) are not crucial for rapid adjustment of noise
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 1
`
`
`
`level. Using W = 1 in Martin’s algorithm is a perfectly reasonable choice for that
`
`parameter.
`
`5. Martin says the overall window length L must be large enough to
`
`bridge any peak of speech activity, but short enough to follow non-stationary noise
`
`variations. He does not make similar comments about the number of sub-windows
`
`W. [Ex. 1006 at 1094.] Martin does not specify any upper or lower bounds on W.
`
`Where W = 1, the length of each sub-window M is equal to the length of the
`
`window L. Martin does not suggest that these values would not work. On the
`
`contrary, he specifies these values as configurable parameters which one in the art
`
`would understand how to set. One in the art would understand that Martin’s
`
`algorithm functions equally well for any positive integer W. [Ex. 1006, Figure 2.]
`
`6. Martin says a window time length of 0.625 (seconds) is “a good
`
`value”, and this value corresponds in his example to window sample length of L =
`
`5000. He does not provide any qualitative assessment of how many sub-
`
`window(s) W would be “good.”
`
`7. Martin explains that W is chosen at least in part on the basis of
`
`“computational complexity and delay” [Ex. 1006 at 1094.] The basis for
`
`“computational complexity” used by Martin at time of publication was in 1993.
`
`Six years later in 1999, at the time of the filing of the ’345 patent, computers were
`
`considerably faster and more capable. Hence a value of W that would be chosen in
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 2
`
`
`
`1993 could differ in 1999 or today. Martin had the foresight to anticipate this
`
`issue, and allow the choice of W to be a design variable, whether smaller or larger.
`
`8. Martin says sub-windows can “improve[e] the noise tracking
`
`capability” for “a rapid noise power increase.” [Ex. 1006 at 1094.] Martin does
`
`not state that this feature is required to track noise. Just that it improves noise
`
`tracking in some circumstances.
`
`9.
`
`According to Martin’s algorithm, the noise floor Pn(i) adjusts to rapid
`
`noise power decreases, because the noise floor is immediately updated if the
`
`current smoothed power is less than the floor. This is true for any W.
`
`10. Other aspects of Martin’s algorithm are not affected by the choice of
`
`W. For example, the noise floor Pn(i) is never allowed to be above 𝑃𝑃�x(i). [Ex.
`1006 at Figure 2.], and 𝑃𝑃�x(i) has no dependency on W.
`
`11. Dr. Douglas asserts that Martin does not update the noise floor Pn(i)
`
`periodically. [Ex. 2002 (-626 Douglas) at ¶72.] That statement is not correct.
`
`Where W = 1, Martin’s noise floor Pn (i) is always set to PMmin at the end of the
`
`window. In Martin’s Figure 2, both branches of the monotonically increasing test
`
`simplify to the same result when W = 1. In the “no” branch, Martin selects the
`
`minimum of the last W values stored in min_vec (italicized for ease of reading).
`
`Martin shows this through the mathematical statement min(min_vec(r*M),
`
`min_vec((r-1)*M),… min_vec(r-W+1)*M). The last value in this list is min_vec(r-
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 3
`
`
`
`W+1)*M, which is equal to min_vec(r-1+1)*M because W = 1, which then
`
`simplifies to min_vec(r*M). Therefore Martin selects min(min_vec(r*M),
`
`min_vec(r*M)), which simplifies to min_vec(r*M). Martin sets min_vec(r*M) =
`
`PMmin, and therefore, the “no” branch simplifies to
`
`Pn(i) = min(min_vec(r*M)) = min(PMmin) = PMmin.
`
`Because the sub-window and the window are the same length, the minimum of
`
`each sub-window is also the minimum of the window.
`
`12. Dr. Douglas asserts that without the “monotonically increasing
`
`power” block, the noise floor Pn(i) would never increase because “the algorithm
`
`would only adjust the noise level downward.” [Ex. 2002 (-626 Douglas) at ¶97.]
`
`But that is not correct. Martin explains that the noise floor would increase after
`
`L+M samples, but where sub-windows are used, “this delay is reduced to M
`
`samples.” [Ex. 1006 at p. 1094.] While the delay is shorter when sub-windows are
`
`used, the noise floor can still increase without sub-windows. In any event, by
`
`choosing W = 1, we may have Pn(i) increase at intervals of M samples.
`
`13. Dr. Douglas’s assertion that the monotonically increasing block of
`
`Martin is “crucial” is also inconsistent with Prof. Rainer Martin’s follow-on work.
`
`In 1994, Prof. Martin published another article that describes use of his noise floor
`
`algorithm in a spectral subtraction system. While Prof. Martin uses sub-windows
`
`(where D is the window length, M is the sub-window length, and W is the number
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 4
`
`
`
`of sub-windows, D = M * W), he does not include a monotonically increasing
`
`determination in that algorithm. [Ex. 1031 (Martin 94) at p. 1183.] Prof. Martin
`
`states that “In case of increasing noise power the update of noise estimates is
`
`delayed by D + M samples.” [Ex. 1031 (Martin 94) at p. 1183.] This is the same
`
`duration described in Martin 93 [Ex. 1006] for the non-monotonically increasing
`
`block. If the monotonically increasing block of Martin 93 were truly a “crucial”
`
`part of the algorithm, Prof. Martin would have included it in his 1994 algorithm.
`
`Prof. Martin’s 1994 paper confirms that a person working in the field would have
`
`viewed that block as an optional feature of the noise floor algorithm.
`
`B. Martin Discloses the Claimed “Future Minimum” Even Where
`Sub-Windows Are Used
`14. Andrea and Dr. Douglas contend that where W = 4, Martin does not
`
`disclose the current and future minima. [Ex. 2002 (-626 Douglas) at ¶¶64-71.]
`
`They are wrong. Even where W = 4 Martin discloses those elements.
`
`15. As noted in my earlier declaration [Ex. 1003, ¶¶136-39], when W = 1
`
`we may map Martin’s Pn(i) to the ’345 “current minimum value” and Martin’s
`
`PMmin to the ’345 “future minimum value”. This discloses the claimed current and
`
`future minima.
`
`16. When W = 4, we may keep this mapping the same. Martin stores each
`
`PMmin value in the min_vec array, and then selects either the most recent or the
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 5
`
`
`
`smallest PMmin value from the past W values for assignment to Pn(i). [Ex. 1006,
`
`1094.] This mapping can, in fact, be applied for any W, and does not contradict the
`
`mapping I give in the paragraph above when W = 1 because
`
`min(min_vec(r*M),…,min_vec((r-W+1)*M)) = PMmin when W = 1.
`
`17. Dr. Douglas addresses the monotonically increasing and non-
`
`monotonically increasing scenarios separately. I will do the same.
`
`18. Dr. Douglas states that where the signal is monotonically increasing,
`
`the noise floor Pn(i) is set equal to the PMmin value of the current sub-window
`
`instead of to the smallest PMmin value across the data window L. [Ex. 2002 (-626
`
`Douglas) at ¶66.] According to Dr. Douglas, this means PMmin cannot be the
`
`“future minimum” because it is not the minimum of the data window L. But
`
`nothing in the claim specifies a particular data window over which the “future
`
`minimum” must be calculated. PMmin is, in fact, a minimum value. Specifically, it
`
`is the minimum value within the sub-window, and that is sufficient to meet the
`
`claim language.
`
`19. Dr. Douglas states that where the signal is not monotonically
`
`increasing, the noise floor Pn(i) is set to a value from min_vec and not to a PMmin
`
`value. [Ex. 2002 (-626 Douglas) at ¶70.] Dr. Douglas mischaracterizes the Martin
`
`algorithm. The min_vec variable stores the PMmin values. This can be seen in the
`
`top box of the excerpt from Figure 2 of Martin (see Figure A on next page). At the
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 6
`
`
`
`end of each sub-window of M samples, Martin stores the PMmin value in min_vec.
`
`Hence, every value in min_vec represents a PMmin value from a sub-window.
`
`
`
`
`
`Figure A
`
`The left hand side of Figure A shows the non-monotonically increasing branch. In
`
`this branch, Martin’s algorithm sets Pn(i) equal to the smallest PMmin value from the
`
`past W sub-windows by selecting the smallest value out of min_vec. Martin’s use
`
`of min_vec as intermediate storage of the PMmin values does not change the fact that
`
`Pn(i) is set equal to the smallest PMmin value in the past W sub-windows.
`
`20. Hence, whether the signal is monotonically increasing or not, Pn(i)
`
`(current minimum) is set equal to a PMmin value (future minimum) at the end of
`
`every sub-window.
`
`
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 7
`
`
`
`
`
`In my first declaration I provided the annotated figure (excerpt reproduced above),
`
`and explained that where W = 1, both boxes (which are in the bottom blue
`
`rectangle) resulted in Pn(i) being set to PMmin. [Ex. 1003 at ¶140.] The same is also
`
`true for W > 1 (e.g., W = 4). Either way, Pn(i) will be set to a PMmin value. If
`
`monotonically increasing is determined in the decision diamond, Pn(i) is set to the
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 8
`
`
`
`most recent PMmin. If not monotonically increasing, Pn(i) is set to the smallest of
`
`the W past PMmin values.
`
`C. Dr. Douglas Misinterprets Martin’s Own Description of How
`Well His Noise Floor Algorithm Works
`21. Andrea and Dr. Douglas contend that Martin said his algorithm is
`
`“biased” and does not work well. [Ex. 2002 (-626 Douglas) at ¶87.]
`
`22. They are wrong because: (i) Martin states his algorithm when used
`
`with spectral subtraction reduces noise by 10 dB without creating musical tones,
`
`which a person skilled in the art would have recognized is a non-trivial
`
`improvement; and (ii) Martin uses the word “bias” to refer to the difference
`
`between the noise floor and the average noise level, and there are ways to
`
`compensate for this effect.
`
`23. Martin explains [Ex. 1006 at p. 1096] that his algorithm was used with
`
`a filter bank in a spectral subtraction experiment. He explains that “few annoying
`
`musical tones” and “improvement of about 10 dB” were both observed. He also
`
`explains [Ex. 1006 at p. 1095] “our SNR [signal-to-noise ratio] estimate shows
`
`good agreement with the true SNR” and his algorithm was “successfully
`
`incorporated in several speech processing systems.” Martin therefore gives ample
`
`incentive for a reader to utilize and build upon his results.
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 9
`
`
`
`24.
`
`In particular, an improvement of 10 dB represents a factor of 10
`
`improvement. Improving the signal-to-noise ratio by 10 dB is equivalent to
`
`reducing the noise power by a factor of ten relative to the signal; this would be
`
`recognized to a person skilled in the art as being very compelling and significant.
`
`One in the art would have understood that noise suppression techniques involved
`
`tradeoffs between noise suppression and the introduction of artifacts into the
`
`cleaned signal. That Martin reports that his approach resulted in few annoying
`
`tones would have been understood as a significant result showing that Martin’s
`
`algorithm works well.
`
`25. Martin explains [Ex. 1006 p. 1095] that “the estimate is biased when
`
`no speech is present.” Martin explains that “the minimum power estimate is
`
`smaller than the true noise power.” [Ex. 1006 p. 1094] By underestimating the
`
`noise in the absence of speech, the algorithm tends to overestimate the signal-to-
`
`noise ratio in the absence of speech [Ex. 1006 Figure 3]. (A smaller denominator
`
`in the signal-to-noise ratio gives a larger signal-to-noise ratio.)
`
`26. Bias is well known to a person skilled in the art as a form of
`
`systematic offset where your estimate differs, on average, from the true value you
`
`are estimating. It is also known to a person skilled in the art that if the bias is
`
`known or can be estimated, then compensation factors can eliminate the bias.
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 10
`
`
`
`27. Martin explains that one way to compensate for the bias is to use a
`
`compensating overestimation factor “ofactor” [Ex. 1006 at p. 1094]. After use of
`
`this factor, Martin explains that “the noise power estimate is approximately
`
`unbiased” [Ex. 1006 at p. 1095]. Hence, the effect of the bias is mitigated and is of
`
`little consequence in Martin.
`
`28. Rather than teach away from Martin, Hirsch also uses adopts an
`
`overestimation factor β in the range of 1.5 to 2.5 that plays a similar role as
`
`“ofactor” in Martin to remove the bias in the noise estimate. A person skilled in
`
`the art would therefore see similarities in the teachings of Hirsch and Martin and
`
`would be motivated to consider the two together.
`
`D. Dr. Douglas Mischaracterizes My Deposition Testimony
`29. Andrea and Dr. Douglas mischaracterize my deposition testimony
`
`about the monotonically increasing block in Martin. [Ex. 2002 (-626 Douglas) at
`
`¶¶94-103.]
`
`30. As I explained during the deposition, a person skilled in the art would
`
`have known there were multiple ways to implement a test to determine whether a
`
`sequence of numbers is increasing. [Ex. 2005 at 69-71.] The details how one
`
`would implement that test vary. Martin does not disclose the particular algorithm
`
`he uses to test for monotonicity, but implementing such a feature was well within
`
`the skill of one in the art. For example, this test could be performed digitally by
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 11
`
`
`
`computer using a sequence of comparisons on min_vec, or a sequence of
`
`comparisons on 𝑃𝑃�x(i).
`
`31. As I explained at my deposition, the implementation details of this
`
`determination are irrelevant to my opinions. See also [Ex. 2005 at pp. 80-85].
`
`E. Dr. Douglas Is Incorrect that Hirsch Teaches That 0.2 Seconds Is
`a Long Time
`32. Dr. Douglas asserts that a person skilled in the art would have viewed
`
`Martin’s 0.625 second period as a long time compared to Hirsch’s 0.4 second
`
`period [-626 Douglas at ¶¶84-85]. He also states that a person skilled in the art
`
`would be discouraged from using Martin due to Hirsch’s statement that most
`
`known approaches require relatively long segments of past speech. He is incorrect.
`
`33. Hirsch cites to several references that describe algorithms for
`
`estimating noise. These include Martin [6], a prior article by Hirsch [7], and an
`
`article by Campernolle [3]. [Ex. 1005 at 153.] In his own prior article, Hirsch
`
`points out that a variety of windows 0.250, 0.5, 1.0, and 2.0 seconds [Ex. 1032 at p.
`
`10] can be considered for estimating noise characteristics in speech. There is a
`
`tradeoff between accuracy of estimating the statistical characteristics of the noise,
`
`and the ability to adapt to the noise. The optimum choice of window depends on
`
`the nature of the speech signals being considered. Hirsch concludes that “a length
`
`of 500 ms [0.5 second] seems to be a good compromise for the cases we have
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 12
`
`
`
`examined” [Ex. 1032 at p. 24]. Hirsch acknowledges that there is no universal
`
`value that is optimal for all cases.
`
`34. Other known approaches to estimating noise operate on segments of
`
`speech that are longer than 0.400 or 0.625 seconds. For example, Helf describes in
`
`[Ex. 1010 at 8:19] that his effective window length is a 10 second interval over
`
`which he performs a search for a minimum.
`
`35. The Campernolle article describes a system where “histograms are
`
`restarted every 10 secs of (presumed) speech” [Ex. 1033 at 14], and “effective time
`
`constants of 20 to 50 seconds depending on the mixing ratio of speech and silence”
`
`[Ex. 1033 at 15].
`
`36. Hence, Martin’s window of 0.625 seconds is perfectly reasonable for
`
`the cases considered by Martin and not significantly different from the 0.400
`
`seconds used by Hirsch.
`
`37. There is no universal agreement on what is the window length that
`
`optimizes the tradeoff on accuracy of noise estimation and the ability to adapt.
`
`Hirsch’s segment length of 0.400 seconds, and Martin’s window length of 0.625
`
`seconds are both reasonable choices, depending on the situation. One does not
`
`preclude the other, and one working in the field would have understood how to
`
`combine the algorithms and select an appropriate time window.
`
`
`
`
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 13
`
`
`
`
`
`* * *
`
`I, Bertrand Hochwald, do hereby declare and state, that all statements made
`
`herein of my own knowledge are true and that all statements made on information
`
`and belief are believed to be true; and further that these statements were made with
`
`the knowledge that willful false statements and the like so made are punishable by
`
`fine or imprisonment, under Section 1001 of Title 18 of the United States Code.
`
`
`
`Executed on: February 7, 2018
`
`
`
`
`
`
`
`
`
`
`
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 14
`
`
`
`Exhibit List
`
`I set forth the list of exhibits that I relied upon in forming my opinions in my
`
`opening declarations Exhibits 1003 and 1004. In addition to those exhibits, I also
`
`have considered the exhibits and documents listed below.
`
`
`
`1027
`
`1028
`
`1029
`
`Exhibit # Reference Name
`
`Institution Decision
`
`Andrea’s Patent Owner Responses
`2002
`Dr. Douglas’s Declaration
`2005
`My Deposition Transcript, dated Oct. 12, 2017
`1026
`Transcript from Deposition of Scott Douglas dated Jan. 17,
`2018
`Exhibit 1 from Douglas Dep., Figure 27 depicting Current and
`Future Minima
`Exhibit 2 from Douglas Dep., Dr. Douglas’s mark up of Exhibit
`1
`Exhibit 8 from Douglas Dep., Declaration of Scott Douglas in
`Support of Complainant Andrea’s Claim Construction Brief in
`Inv. No. 337-TA-949 (Oct. 19, 2015)
`Transcript from Deposition of Scott Douglas dated June 16,
`2017, taken in Inv. No. 337-TA-1026
`Rainer Martin, Spectral Subtraction Based on Minimum
`Statistics, Proc. EUSIPCO-94, pp. 1182-85 (1994) (“Martin
`94”)
`H. G. Hirsch, “Estimation of Noise Spectrum and its
`Application to SNR Estimation and Speech Enhancement,”
`Technical Report TR-93-012, International Computer Science
`Institute (1993) (reference [7] in Martin 93)
`D. Van Campernolle, “Noise Adaptation in a Hidden Markov
`Model Speech Recognition System”, Computer Speech and
`
`1030
`
`1031
`
`1032
`
`1033
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 15
`
`
`
`Language, Vol. 3, pp. 151-167 (1989) (reference [3] in Martin
`93)
`
`
`
`
`
`
`
`Reply Declaration of Dr. Bertrand Hochwald
`
`IPR No. 2017-00626
`Apple Inc. v. Andrea Electronics Inc. - Ex. 1023, p. 16
`
`
Accessing this document will incur an additional charge of $.
After purchase, you can access this document again without charge.
Accept $ Charge
Still Working On It
This document is taking longer than usual to download. This can happen if we need to contact the court directly to obtain the document and their servers are running slowly.
Give it another minute or two to complete, and then try the refresh button.
A few More Minutes ... Still Working
It can take up to 5 minutes for us to download a document if the court servers are running slowly.
Thank you for your continued patience.
This document could not be displayed.
We could not find this document within its docket. Please go back to the docket page and check the link. If that does not work, go back to the docket and refresh it to pull the newest information.
Your account does not support viewing this document.
You need a Paid Account to view this document. Click here to change your account type.
Your account does not support viewing this document.
Set your membership
status to view this document.
With a Docket Alarm membership, you'll
get a whole lot more, including:
- Up-to-date information for this case.
- Email alerts whenever there is an update.
- Full text search for other cases.
- Get email alerts whenever a new case matches your search.
One Moment Please
The filing “” is large (MB) and is being downloaded.
Please refresh this page in a few minutes to see if the filing has been downloaded. The filing will also be emailed to you when the download completes.
Your document is on its way!
If you do not receive the document in five minutes, contact support at support@docketalarm.com.
Sealed Document
We are unable to display this document, it may be under a court ordered seal.
If you have proper credentials to access the file, you may proceed directly to the court's system using your government issued username and password.
Access Government Site
