`
`
`
`
`
`
` Exhibit 6
`
`
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 2 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 1
`
` UNITED STATES DISTRICT COURT
` SOUTHERN DISTRICT OF NEW YORK
`___________________________________
` )
`NETWORK-1 TECHNOLOGIES, INC., )
` )
` Plaintiff, )
` )
`vs. ) No. 14 Civ. 2396
` ) 14 Civ. 9558
` ) (PGG)
`GOOGLE LLC and YOUTUBE, LLC, )
` )
` Defendants. )
`___________________________________)
`
` CONFIDENTIAL ATTORNEYS' EYES ONLY
` VIDEOTAPED DEPOSITION OF AUDIBLE MAGIC 30(b)(6)
` ERLING WOLD
` September 4, 2019 at 10:01 a.m.
` Three Embarcadero Center, 26th Floor
` San Francisco, California
`
`REPORTED BY: LANA L. LOPER,
` RMR, CRR, CCP, CME, CLR, CSR No. 9667
`____________________________________________________
` DIGITAL EVIDENCE GROUP
` 1730 M Street, NW, Suite 812
` Washington, D.C. 20036
` (202) 232-0646
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 3 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 14
`
`C-l-a-n-g-o?
` A Yes.
` Q And in general terms, what was Clango?
` A Clango was intended for people to be listening
`to, say, an Internet radio station. And when they were
`listening to an Internet radio station, they can run
`this other operation called Clango on their desktop.
` And if they heard a tune that they liked, they
`could press a button, I think, and they would identify
`that tune that was playing and then allow you to, you
`know, purchase it, for example, if you were interested.
` It showed you the metadata, and then it gave
`you a URL to an e-commerce site.
` Q Is it fair to say that Clango identified songs
`by their content?
` A Yes.
` Q Why try to identify songs by -- by their
`content?
` A Well, at the time there wasn't -- there wasn't
`always metadata. So if -- if someone was just listening
`to a radio station on the Internet, it might just be a
`stream from a terrestrial radio station, and so it
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 4 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 29
`
`simple idea.
` It's basically, at the time we were using
`10 MFCCs per segment of sound. And a segment of sound,
`you know, could be 25 milliseconds, or it could be a
`second, for example, something in that -- but it's
`something short, in that range.
` And the -- the way those -- so they're just
`floating by numbers.
` Let me think.
` The way that they are computed is, you do some
`kind of spectral analysis of the sound. So that's,
`like, what the ear does. You look for low frequencies,
`high frequencies, so on. And then you rectify that.
`And then you take another spectral analysis of that.
` And -- and what you end up with is a set of
`features that really describes the spectral shape of the
`sound. The lower-order features are kind of the gross
`shape of the spectrum, and the higher-order features are
`more the detail, the -- yeah, the higher features on the
`spectrum, and over time.
` Q And once the Clango application had extracted
`those features, what would it do next in the process of
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 5 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 30
`
`identifying a song?
` MR. LEDAHL: Lacks foundation.
` THE WITNESS: So in the actual application, as
`it was released in mid-2000, there would be the Clango
`application, itself, would just extract the features.
`And so it would -- when the person pressed the button,
`there would be, say, a circular buffer of the last -- or
`not necessarily a circular buffer. There would be a
`buffer containing the last so many seconds of the sound
`of what they had been listening to.
` That would all be fingerprinted at the Clango
`application. So this set of MFCCs would be extracted.
` Then that would -- that package would be sent
`over the network to a server at -- under Audible Magic's
`control. I think they actually had it in a third-party
`site at the time. I'm not sure.
` And at that -- and on -- on that end, there
`would be a reference database of fingerprints. And then
`some kind of lookup algorithm would be done to compare
`the fingerprint coming in with that set of references
`and -- to see if any of them were close enough that we
`would report a match.
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 6 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 93
` Q And how were those fingerprints indexed for
`lookup as of the July 2000 alpha Clango release?
` A So the -- in the July release, we had moved
`from the -- the tree that I had mentioned earlier in my
`deposition, to a k-d tree, which is just another way of
`segmenting up the reference space, future space. It's
`a -- it's a binary tree.
` Q And how did searches of query fingerprints in
`the July 2000 k-d tree proceed?
` A So you would first -- um, well, a tree would
`have to be built in the -- in the database. So that --
`that was a step that was taken.
` So, you know, you would read in this database,
`and you would do an indexing step. The tree would be
`constructed. So that was before any queries were done.
` But when a query was done, then it would follow
`that path in a similar way to the way I mentioned
`before. Since this was the binary tree, you would
`just -- at each layer of the tree, you would be
`splitting -- you would be choosing one half of the
`remaining section of the possibilities, until you got to
`a leaf. And then you would compare it to everything
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 7 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 94
`
`there.
` Q Did searches of a query fingerprint in the July
`2000 k-d tree compare that fingerprint to every
`reference fingerprint in the k-d tree?
` A No, not typically.
` Q Did the search of the k-d tree only identify
`exact matches of the reference -- of the query
`fingerprint?
` A Well, again, I -- it depends what you mean by
`exact.
` No, the -- typically, you would -- you would
`find -- I mean, actually, I should just say no.
` It would not only find exact matches; it would
`find similar enough matches.
` Q How did the search algorithm decide whether a
`match was similar enough?
` A So the -- after you followed this tree down to
`the bottom, then you would compare the query to every
`reference in that leaf. And that was a Euclidean
`distance at the time.
` So we were just taking the set of however many
`numbers, 200 number, so 200 dimensional feature vector,
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 8 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 98
`linearly, so less than order of N, where N is the
`database size.
`BY MR. DANG:
` Q So to clarify, it scaled less than linearly
`with the size of the database?
` A Correct.
` Q Was that true even if the computer processing
`power was held constant?
` A Yes.
` Q As of the August 2000 beta release, how were
`the reference fingerprints indexed in the back-end
`database?
` A Sorry, say that once more time. I was just
`thinking to the answer to the question before, so
`please?
` Q No problem.
` As of the August 2000 beta release --
` A Yes.
` Q -- how were the reference fingerprints indexed
`in the back-end database?
` A Yeah, so that was using this k-d tree that I
`mentioned.
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 9 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 204
`
`ID server at the back end.
` Q Now, the actual search, if you will, of the
`database, that was run on Audible Magic servers,
`correct?
` A Yes.
` It's -- it's possible that, even at that time,
`there may have been third-party server companies or
`something like that. But they were our servers, yes.
` Q Well -- well, let me -- let me see if I can
`differentiate a little more clearly.
` So Clango was essentially a client application
`that would be downloaded onto a user's computer, right?
` A That's correct.
` Q But the Clango application that a client would
`download, that didn't actually run the search, did it?
` A That is correct.
` Q The search was on something else that the user
`didn't have access to?
` A That's correct, except through the sending of
`the package to it, yes.
` Q But it was a black box the user had no
`understanding of, correct?
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 10 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 205
`
` A Right.
` MR. DANG: Object to the form.
` THE WITNESS: I mean, some users may have known
`something, but yes.
`BY MR. LEDAHL:
` Q But it wasn't -- it's not as though code for
`running the search was downloaded onto a user's
`computer. Is that right?
` A Correct.
` Q And it's not --
` A Actually, you know what, I'm not sure about
`that. I actually wonder if we didn't distinguish,
`because we -- we gave them one piece of code. It's --
`it's possible that that code existed on the Clango
`application. I -- I would have to check into that.
` Q You just don't know?
` A But it wasn't called directly, if that's the
`distinction you're trying to make.
` Q Well, was there -- I -- I guess users didn't
`have -- I mean, for one thing, they didn't have the
`database, right?
` A Correct.
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 11 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 212
` a high hit rate on the tier 1 database.
` Does that help you understand the distinction?
` A That -- that makes me a little bit unclear what
`it is, because the daily updates makes it sound like
`it's the first thing I said; that the tier 2 database
`was to make -- now, maybe -- actually, now that I'm
`rereading it...
` Ah, now that I'm rereading it, actually, I
`think now it makes more sense.
` I think that the tier 2 database was then
`incorporated into the tier 1 database on a daily basis.
` Q Was -- was there ever a -- a database of what I
`will call more frequently hit references?
` A Later, but not at this time.
` Q Now, what was the first time that you or
`anybody at Audible Magic made any public disclosure
`about any of the indexing techniques you use there?
` MR. RAMSEY: Objection. Vague.
` THE WITNESS: Yeah, I mean, that's a pretty
`broad question.
` I mean, in some ways, we've never made public
`disclosure. We -- except I -- I -- I mean, it depends
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 12 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 213
`
`on what you mean.
` I mean, obviously, the -- the product exists in
`a sense that it's a public disclosure. And I'm sure
`there have been casual mentions to customers or
`important suppliers or other people, but I -- I don't
`know. I mean, that's a really big question.
`BY MR. LEDAHL:
` Q Now, earlier, you talked about when -- the idea
`that a potential user could download the -- the
`demonstration or the -- the alpha version of the Clango
`application client, right?
` A Right.
` Q And I think you said -- but I want to make sure
`I understood correctly -- that there was some kind of
`end-user license agreement, but you didn't know what the
`terms of that were?
` A This is a vague memory that -- in the initial
`version, there was some kind of end-user license
`agreement. And, no, I do not.
` If -- if that's true, I don't remember the
`terms.
` Q And is the same true for the beta, that you
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 13 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 297
` MR. RAMSEY: I want to designate the entire
`thing confidential, attorney eyes only. And we'll all
`work in good faith as we proceed to de-designate, if
`necessary.
` THE REPORTER: Read and sign?
` MR. RAMSEY: Yes.
` (END TIME: 4:39 p.m.)
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8 9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 234-8 Filed 11/11/20 Page 14 of 14
`
`9/4/2019
`
`Network-1 Technologies, v. Google LLC and Youtube LLC
`Confidential - Attorneys' Eyes Only
`
`Erling Wold 30(b)(6)
`
`Page 298
`
`STATE OF CALIFORNIA )
`COUNTY OF SAN FRANCISCO )
`
` I, Lana L. Loper, a Certified Shorthand
`Reporter, do hereby certify:
` That prior to being examined, the witness in
`the foregoing proceedings was by me duly sworn to
`testify to the truth, the whole truth, and nothing but
`the truth;
` That said proceedings were taken before me at
`the time and place therein set forth and were taken down
`by me in shorthand and thereafter transcribed into
`typewriting under my direction and supervision;
` I further certify that, pursuant to FCRP Rule
`30(e)(1), the review and signature by the witness:
`
` __X_ was requested by the deponent or a
` party before the conclusion of the
` deposition;
` ____ was not requested by the deponent
` or a party before the conclusion of the
` deposition;
`
` I further certify that I am neither counsel
`for, nor related to, any party to said proceedings, nor
`in anywise interested in the outcome thereof.
` In witness whereof, I have hereunto subscribed
`my name.
`
`Dated: September 17, 2019
`
`___________________________________________________
`LANA L. LOPER, RMR, CRR, CCP, CME, CLR, CSR 9667
`
`1
`2
`
`3 4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`11
`
`12
`13
`
`14
`15
`16
`
`17
`18
`
`19
`20
`21
`
`22
`
`www.DigitalEvidenceGroup.com
`
`Digital Evidence Group C'rt 2019
`
`202-232-0646
`
`