1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Page 193
`
`embodiment of Conwell that does the neighbor search,
`we've got the Element (b), those digitally created
`compact electronic representations of a first
`electronic work, because we're taking a robust hash
`value of our unknown work?
` A Right. So just to give a specific
`example, you have an original song; you apply this
`hashing -- this robust hashing algorithm; you obtain
`a 128-bit sequence, which translates to Number 198.
`Okay? So it goes in your database.
` Then you have a version of that song which
`is not identical. It could be, say, MP3 recorded at
`a different sampling rate. You apply -- you go
`through this procedure; you apply a robust hash
`algorithm, and if indeed, you know, that song was
`similar, with very high probability the hash values
`will be the same; and therefore, we obtain, again,
`198.
` So then we simply do a lookup in the
`table, and we have two songs that are simply a
`version of each other, and because they map to the
`exact same identifier, 198, they are deemed to -- to
`be similar, and so they are in the neighborhood of
` NETWORK-1 EXHIBIT A2006
`each other.
` Google Inc. v. Network-1 Technologies, Inc.
` Q Element (c) says we're going to be
` IPR2015-00345
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 193 of 384
`
`

`
`Page 194
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`10
`11
`11
`12
`12
`13
`13
`14
`14
`15
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`comparing the first electronic data with the second
`digitally created compact electronic representation
`using a nonexhaustive neighbor search.
` Do you see that?
` A Yes. That's 13(c); right?
` Q Yes.
` A Yes.
` Q Is that, as you understand, Conwell --
`withdrawn.
` Is it your testimony that Conwell teaches
`Is it your testimony that Conwell teaches
`Element (c) when it teaches using a robust hash
`Element (c) when it teaches using a robust hash
`approach, where it uses a lookup table to compare
`approach, where it uses a lookup table to compare
`one hash value of the unknown work to hash values
`one hash value of the unknown work to hash values
`that are in the database?
`that are in the database?
` A Yes.
`A
`Yes.
` Q In the embodiment in Conwell that --
`withdrawn.
` Would you agree that Conwell teaches some
`things are not a neighbor search?
` A I will have to read the entire patent. I
`mean, clearly, the intent is to map similar songs to
`the same identifier. So that is the whole point of
`the patent.
` Q Well, if -- does Conwell teach using a
`nonrobust hash?
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 194 of 384
`
`

`
`Page 195
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
` In other words, does it say that we can
`use a nonrobust hash, or does it say the only thing
`you can ever use in this is a robust hash?
` A We have to read the entire patent.
`Clearly, the intent is to use -- it teaches,
`actually, using a robust hash. It teaches that.
` Q That's one of the embodiments; right?
` A It is the -- the main one. That's my
`opinion --
` Q All right. The main one.
` A Huh?
` Q It's the main one?
` A Yes.
` Q But it teaches other things as well;
`right?
` A It teaches many things, yes.
` Q Now, does it teach some things that would
`not be a neighbor search?
` A I would have to read the whole patent
`again. I focused on the neighbor search.
` Q By "focused on the neighbor search," you
`mean you focused on the embodiment that uses the
`robust hash; right?
` A Yes.
` Q All right. Then let me see how that maps
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 195 of 384
`
`

`
`Page 196
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`onto the claim, because I'm not following you.
` If we look at Claim 13, we've got a
`database that includes various things.
` That's Element (a); right?
` A Yes.
` Q Among those are "First electronic data,
`including a first digitally created compact
`electronic representation of one or more referenced
`electronic works"; right?
` A Yes.
` Q In Conwell in the embodiment you're
`pointing to, the "electronic works" are various
`digital songs, right, in a database?
` A Yes.
` Q In Conwell in the embodiment you're
`pointing to, is the "compact electronic
`representation" the hash value?
` A Yes, it is.
` Q Now, the next element, (a)(2), is
`"electronic data related to an action."
` Do you see that?
` A Yes.
` Q And it relates to an advertisement.
` Do you see that?
` A Yes.
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 196 of 384
`
`

`
`Page 197
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
` Q Then Element (b) says, "Obtaining a second
`digitally created compact electronic representation
`of a first electronic work."
` A Yes.
` Q In Conwell in the embodiment you're
`pointing to, what constitutes the second digitally
`created compact electronic representation?
` A So you -- you have the -- what's called
`the query song, okay, and so that will an electronic
`work. And you extract a hash from it, a robust
`hash, and you obtain a compact electronic
`representation of that work.
` Q Is the robust hash value the digitally
`created compact electronic representation?
` A It is, yes.
` MR. ELACQUA: Is this a good breaking point?
` MR. DOVEL: Yeah. Let's do it.
` THE VIDEOGRAPHER: We're off the record at
`2:42 PM.
` (Off the record.)
` THE VIDEOGRAPHER: We are back on the record
`at 2:54 PM.
`BY MR. DOVEL:
` Q Let's continue with Element (c) of
`Claim 13, which is on page 50. It carries over to
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 197 of 384
`
`

`
`Page 198
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`page 51 of your Declaration.
` In the embodiment in Conwell that you
`assert discloses the neighbor search -- withdrawn.
` The claim requires in Element (c) that
`we're going to compare the first electronic data
`with the second digitally created compact electronic
`representation using a nonexhaustive neighbor
`search.
` That's part of it; right?
` A Yes.
` Q In your -- in the embodiment in Conwell
`that you assert discloses this claim limitation,
`what is it that constitutes the second digitally
`created compact electronic representation?
` Is it still the hash value from the hash
`of the unknown work using the robust hash, or does
`it change and become something else now?
` A It is comparing the hash values.
` Q Then, when it says "using a nonexhaustive
`neighbor search" in the embodiment in Conwell you've
`identified as meeting this claim, is it -- what is
`it that constitutes comparing using the
`nonexhaustive neighbor search? Is that looking up
`the values in the table?
` A Yes.
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 198 of 384
`
`

`
`Page 199
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`18
`19
`19
`20
`20
`21
`21
`22
`23
`23
`24
`24
`25
`25
`
` Q Is the search that's performed --
`withdrawn.
` Now, when you say "looking up something in
`a table," how can you say that looking up something
`in a table is a search?
` A Because the step that led to the
`identifier on the left of the table, as we just
`discussed, means that you compute an identifier.
`And so your search either has an outcome, which, in
`this case, is an entry in the lookup table, or may
`not have an outcome because the item cannot be
`found.
` Q When you say it's a search, why isn't --
`isn't a search something where you've got to look
`over -- through a wide variety of data?
` A Yeah. But it could be very simple in some
`problems, as here.
` Q Are you saying a lookup table constitutes
`Q
`Are you saying a lookup table constitutes
`a search?
`a search?
` A You can do a search from a lookup table.
`A
`You can do a search from a lookup table.
`It's very easy.
`It's very easy.
` Q Is it the case that -- withdrawn.
` Is it your assertion that the
`Is it your assertion that the
`nonexhaustive neighbor search here is the search
`nonexhaustive neighbor search here is the search
`that's done when we use the hash value lookup table?
`that's done when we use the hash value lookup table?
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 199 of 384
`
`

`
`Page 200
`
`
`11
`2
`2
`3
`3
`4
`4
`5
`5
`6
`7
`8
`9
`10
`11
`11
`12
`12
`13
`13
`14
`14
`15
`15
`16
`16
`17
`17
`18
`18
`19
`19
`20
`20
`21
`21
`22
`22
`23
`23
`24
`24
`25
`25
`
`
` A Yes. So if you have computed -- say ourA Yes. So if you have computed -- say our
`
`identifier was 198 in decimal representation, we
`identifier was 198 in decimal representation, we
`simply have to look up, is that number in the left
`simply have to look up, is that number in the left
`column of the table. And that's a straightforward
`column of the table. And that's a straightforward
`thing to do; so the search is trivial.
`thing to do; so the search is trivial.
` Q In the -- is that the only search that's
`disclosed in Conwell, this lookup table?
` A I would have to read the whole patent.
`The main point is what I just said here.
` Q Is it the case that -- that -- withdrawn.
` If we look at the search you've identified
`If we look at the search you've identified
`as constituting the neighbor search, it's a search
`as constituting the neighbor search, it's a search
`wherein we start with a hash value and then go to a
`wherein we start with a hash value and then go to a
`lookup table to see if it appears in that table; is
`lookup table to see if it appears in that table; is
`that right?
`that right?
` A Yes.
`A
`Yes.
` Q Is the result of that going to be either
`Q
`Is the result of that going to be either
`an exact match of the hash or a determination that
`an exact match of the hash or a determination that
`it does not exist?
`it does not exist?
` A Yeah. It's either the table or it's not.
`A
`Yeah. It's either the table or it's not.
` Q Are we ever going to have a circumstance
`Q
`Are we ever going to have a circumstance
`where we look in the table for that hash value and
`where we look in the table for that hash value and
`we conclude that the hash value does not appear, but
`we conclude that the hash value does not appear, but
`here's one that's pretty close, and we'll return
`here's one that's pretty close, and we'll return
`that one?
`that one?
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 200 of 384
`
`

`
`Page 201
`
`
`11
`2
`2
`3
`4
`4
`5
`5
`6
`6
`7
`7
`8
`8
`9
`9
`10
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`
` A It's not -- it does not -- Conwell doesA It's not -- it does not -- Conwell does
`
`not disclose that.
`not disclose that.
` Q If we return the -- withdrawn.
` So the only search that's disclosed in --
`So the only search that's disclosed in --
`in Conwell using the hash values is one where we
`in Conwell using the hash values is one where we
`look for an exact match, and it's either there or it
`look for an exact match, and it's either there or it
`isn't; is that right?
`isn't; is that right?
`A
`Well, an exact match of that identifier.
` A Well, an exact match of that identifier.
`Like 198, is it there or not? If I have 199, I
`Like 198, is it there or not? If I have 199, I
`would say it's not there.
`would say it's not there.
` Q Is it the case that the Conwell search
`looks for an exact match of the digitally created
`compact electronic representation with one of the
`items that constitutes the first electronic data?
` A So that's understood, when we say "exact
`match," we're talking about the exact match of the
`identifier. Okay? It's not an exact match of the
`works.
` Q When we say "exact match of the
`identifier," in the embodiment you're pointing to in
`Conwell, the identifiers are all the hundred -- are
`all robust hash values; is that right?
` A Yes.
` Q Now let's take a look at Claim 1.
` A In the '179 patent?
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 201 of 384
`
`

`
`Page 202
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
` Q Yes. In Claim 1, we have a database
`comprising first electronic data related to
`identification of one or more referenced electronic
`works.
` In the Conwell embodiment that you
`identify as disclosing this claim, what is it that
`constitutes the first electronic data related to
`identification of one or more referenced electronics
`works?
` A Well, it is illustrated by the table. So
`it is the identifiers that we -- that are
`represented in the left column of Figure 3. These
`are the identifiers derived from content.
` Q Is the -- are the identifiers in the
`embodiment you say anticipates this claim robust
`hash values?
` A That part has nothing to do with
`robustness. So it anticipates it because it
`describes identifiers.
` Q Well, you understand that for the claim as
`a whole, to anticipate it, you've got to use the
`same thing all the way through?
` You can't say, "Well, this embodiment has
`a first electronic data, and this embodiment has a
`search, but the two embodiments are never used
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 202 of 384
`
`

`
`Page 203
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`together."
` You understand?
` A I understand.
` Q All right. Is it the case that in the
`identify you identify that anticipates in Conwell,
`the first electronic data related to identification
`of one or more referenced electronic works is the
`database that includes the robust hash values for
`various works?
` A Yes. If you use robust hash as you
`should, you indeed get -- get that property.
` Q Okay. Let's move down to Element (b).
`"Obtaining by the computer system extracted features
`of a first electronic work."
` In the embodiment that you say anticipates
`the claim, the embodiment of Conwell, what is it
`that constitutes the extracted features of a first
`electronic work?
` A That's obtained by the hashing algorithm.
` Q It's going to be the hash values?
` A Yes.
` Q So if we have an unknown work, in order to
`create -- to extract the feature, what we're going
`to do is use our robust hash to create a hash value?
` A Yes.
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 203 of 384
`
`

`
`Page 204
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
` Q Then in Element (c), it requires
`identifying the first electronic work by comparing
`the extracted features of the first electronic work
`with the first electronic data in the database using
`a nonexhaustive neighbor search.
` Do you see that?
` A Right.
` Q What are the two things that we're
`comparing in the embodiment that you assert -- in
`Conwell that you assert anticipates?
` A So we are comparing the hash strings, or
`equivalently, their decimal representations.
` Q Well, you've identified one thing. Now
`compared two things.
` What are the two things we're comparing?
` A We're comparing two hash values, or it's
`complete equivalent, the hash value can be
`represented in decimal format, as illustrated in
`Figure 3. It's the same. So we compare the two
`hashes or we compare their decimal representations.
` Q When you say "we compare the two," we're
`comparing the hash value of the unknown work with a
`hash value of a known work in the database; is that
`right?
` A Yes. Yes.
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 204 of 384
`
`

`
`Page 205
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
` Q And it may be that we're comparing 128-bit
`representations or decimal representations?
` A It's immaterial.
` Q Either way, it's your assertion that by
`comparing the hash value of a unknown work to a hash
`value of a reference work, we are comparing the
`extracted features of the first electronic work with
`the first electronic data in the database?
` A Yes.
` Q Then the next part is using a
`nonexhaustive neighbor search.
` Is the -- using a nonexhaustive neighbor
`search, is that the -- using the lookup table to see
`whether the hash value of the unknown work appears?
` A Yeah. Using the lookup table is the
`simple way in which the search is implemented.
` MR. DOVEL: Now, is it the case --
` I'm holding a document that has been
`marked as Exhibit 1012. This is the Iwamura patent.
`This will be Exhibit 1012 here.
` (Exhibit 1012 was marked for identification
` by the court reporter and is attached
` hereto.)
`BY MR. DOVEL:
` Q In Iwamura, it teaches a system that
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 205 of 384
`
`

`
`Page 206
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`searches a music database with melody information;
`is that right?
` A Yes.
` Q If we look at the abstract of Iwamura, it
`says, "A remote music database with melody
`information is searched"; is that right?
` A Yes.
` Q In doing that searches -- in doing that
`search, what we're -- again, we're starting with a
`query or an unknown melody; is that right?
` A Yes.
` Q When we do the search that's described in
`Iwamura, is the -- are the -- the possible matches
`the set of the melodies in the database in Iwamura?
` A Yeah, that you try to match melodies.
` Q Now, the search that's -- that's described
`in Iwamura -- in fact, I'll have you look at
`Column 7.
` Now, Iwamura teaches a type of peak note
`search or what it calls a peak search; is that
`right?
` A Among other things, yes.
` Q Well, in addition to teaching a peak
`search, it teaches variations of that where you
`would use dips rather than peaks. That would be one
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 206 of 384
`
`

`
`Page 207
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`18
`19
`19
`20
`20
`21
`21
`22
`22
`23
`23
`24
`25
`
`variation of it?
` A Or both.
` Q But what it -- and then it also says you
`could use the Boyer-Moore algorithm; right?
` A I think I would have to double-check.
`Okay?
` Q Well, whether it does or not. But --
` A Yeah. I think the answer is yes. I just
`want to double-check.
` Q Let's talk -- let's talk about the main
`part.
` What it's mainly focused on here is a
`particular type of what it thinks is -- what Iwamura
`describes as a -- as a new approach or a novel
`search that involves searching with peaks or dips
`and so on; right?
` A Right.
` Q Now, in the type of search that's
`Q
`Now, in the type of search that's
`identified and described in Iwamura, what it does is
`identified and described in Iwamura, what it does is
`it does a comparison of the unknown melody to each
`it does a comparison of the unknown melody to each
`of the melody patterns that are in the melody
`of the melody patterns that are in the melody
`database; right?
`database; right?
` A Yes.
`A
`Yes.
` Q For example, if we look at -- what that
`means, first, is that we are going to take our
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 207 of 384
`
`

`
`Page 208
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`unknown melody, and we're going to use this peak
`search comparison method to see whether it matches
`each of the melodies that are -- that are in the
`database; right?
` A Yes.
` Q Would you agree that it -- that the search
`that's identified -- withdrawn.
` Is it the case that the search identified
`in Iwamura does a comparison of all of the possible
`matches?
` A No.
` Q Why not?
` A Well, because they define what a match is.
`Let me retrieve that.
` Yes. If you look at, actually, Column
`Number 7, say line 34, the criterion is total,
`absolute difference. And by basically taking
`shortcuts, not using all the data, the -- you are
`then unable to evaluate the absolute difference for
`every possible match.
` Q Well, you're saying that in doing the
`comparison, they don't need to look at every bit of
`data in every melody in the reference?
` A That's what they do to save on computation
`time. So it's an approximate search.
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 208 of 384
`
`

`
`Page 209
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
` Q All right. Would you agree that they look
`at all of the melodies that are possible matches?
` A They look at all the songs, yes. But they
`don't use all the data within each song.
` Q Is it the case that the search proceeds --
`withdrawn.
` The -- in Iwamura, is it the case that
`what it identifies as a match would be one of the
`melodies in the database?
` A Yes.
` Q Is the set of all possible matches in the
`Iwamura search the set of all melodies in the
`database?
` A No. Because it does not evaluate, for
`instance, every possible shift. And whenever it
`evaluates a shift, it does not use the matching
`criterion that they ideally would like to use for an
`exact match or an exact search; they use an
`approximation to it.
` Q Is it the case that if we -- withdrawn.
` If a melody is identified as a match using
`the Iwamura algorithm, would the melody then be --
`withdrawn.
` If the Iwamura search approach determines
`that a particular melody meets the matching
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 209 of 384
`
`

`
`Page 210
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`criterion, would it then identify the melody as a
`match?
` A It would declare it as a match.
` Q And when it declares it as a match, what
`does it do? How does it do that?
` A Well, it evaluates the matching criterion
`and then it -- among all the possible potential
`matches it evaluates, it tries to find the best fit;
`right?
` Let's say it wants to find one single
`matching melody, okay, then it would try to find the
`one that minimizes that -- the value of that
`criterion.
` Q It's going to find the one that either --
`it's going to identify the melodies that are either
`exact matches, because the absolute total difference
`is zero, or the melodies that have the least
`absolute difference; is that right?
` A That's right. The -- the perfect match is
`a rare, ideal case. So usually, the criterion will
`not have zero value, and it tries to find the best
`match.
` Q Is it the case that in Iwamura, that the
`match that's returned is an identification of a
`particular melody in the reference database?
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 210 of 384
`
`

`
`Page 211
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
` A Yes. It identifies a song once it
`declares a match.
` Q Does it -- the way the Iwamura system
`operates, does it declare that a part of a song
`matches the melody, or does it declare that the song
`matches or is the best match for the melody?
` A Well, it searches for the song title;
`right? So the melody that's input by the user, you
`know, you try to find a match somewhere in -- in the
`song. It does not really matter where. The
`important thing is to identify the song.
` Q Now, what the -- if we look at Claim 25 of
`the '237 patent -- get that language in front of
`you.
` A Claim 25 of the patent. Oh, it's here.
` Q I'll find it for you in your report.
` A The patent is here unless you want me to
`look in my chart.
` MR. DOVEL: The patent is fine.
` If you look at Claim 25, if you can -- if
`that works for you. Otherwise, I'll find it in your
`chart. Actually, let's don't do 25; let's do -- I'm
`going to give you --
` I'm going to have you look in the
`different report here. I'm going to mark as
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 211 of 384
`
`

`
`Page 212
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Exhibit 1004 -- 1004-988, Dr. Moulin's report
`relating to the '988 patent.
` (Exhibit 1004-988 was marked for
` identification by the court reporter and is
` attached hereto.)
`BY MR. DOVEL:
` Q I'll find the right page for you here.
`Just one second.
` Let's go take a look at page 75.
`Actually, I'll have you look at the claim chart. So
`let's go over to page 80.
` Do you see there's claim language for
`Claim 15 that's reproduced there?
` A On page 80?
` Q Yes.
` A Uh-huh.
` Q In Element 15 -- sorry. Claim -- in
`Claim 15, Element (b) requires doing a nonexhaustive
`search identifying a neighbor.
` Do you see that?
` A Yes.
` Q And you incorporate by reference from
`Claim Element 1(c).
` Do you see that?
` A Yes.
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 212 of 384
`
`

`
`Page 213
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`23
`24
`24
`25
`25
`
` Q And if we go back to page 78 and 79,
`you've got your claim chart that identifies what you
`point to there as identifying as the nonexhaustive
`search; right?
` A Correct.
` Q Okay. I'm going to have you -- let me
`just check. Just one second.
` I'm going to have you -- let's go back to
`your '237 Declaration now. I'd like you to turn to
`page 71. I'm sorry. I should say paragraph 71.
`That's back on page 27.
` Now, in paragraph 71, you present some
`analysis to demonstrate how Iwamura teaches a search
`that can be nonexhaustive.
` Do you see that?
` A Yes.
` Q Here, what you identify as a nonexhaustive
`search is searching that uses peak notes.
` A "Here," you mean on page 27?
` Q Yes. In paragraph 71. We're focused on
`that first.
` A I give that example, yes.
` Q Would you agree that in Iwamura's peak
`Q
`Would you agree that in Iwamura's peak
`search note, that it does search across each melody
`search note, that it does search across each melody
`in the melody database?
`in the melody database?
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 213 of 384
`
`

`
`Page 214
`
`
`11
`2
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`
` A Yes, it does. It's going to compare itA Yes, it does. It's going to compare it
`
`with every song in the database, typically.
`with every song in the database, typically.
` Q In doing the peak search, is it comparing
`the reference -- the unknown work with each of the
`possible melodies that could be returned as a match?
` A It does not evaluate the matching
`criterion. So it discloses using least absolute
`values, and it only computes an approximation to
`that for -- for various reasons. One of them is not
`using all the data. So that's what it does. It's
`only an approximation.
` Q My question was, in doing its comparison,
`is it the case that the Iwamura peak note search
`will do a comparison to each of the possible
`melodies that could be a match?
` A You give only one part of -- of the -- the
`question.
` It does not evaluate all the possible
`melodies because it does not use all the data. As
`soon as you don't use some of the data, you are not
`evaluating every possible melody.
` Q In the database, it identifies -- the
`database consists of a set of known melodies; right?
` A Yes.
` Q And when the Iwamura peak note search
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 214 of 384
`
`

`
`Page 215
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`performs -- before it's run, we've got an unknown
`work in front of us; right?
` A Uh-huh.
` Q And at that point, the possible melodies
`are the melodies that are -- rather, the possible
`matches are all the melodies that are in the
`database; right?
` A I just want to make clear -- okay. A
`melody in the context of Iwamura is a short segment;
`right? And then you have a whole audio piece which
`could be much longer. And so you're comparing that
`short segment with short segments from that audio
`piece.
` You are not necessarily comparing with
`every segment, for instance, or you're not
`necessarily using all the data within each segment.
`But it is true that you do evaluate each audio work
`in the database.
` Q When you say "evaluate," it does a search
`by comparing the unknown work to each of the
`melodies that are in the database; right?
` A I'm making --
` MR. ELACQUA: Objection.
` THE WITNESS: I'm making a distinction, just
`like Iwamura does, between the melody itself and the
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 215 of 384
`
`

`
`Page 216
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`entire piece of music. So the melody that's input
`by the user could be very short, say, a few seconds,
`five seconds. The piece of music in the database
`could be much longer.
` So I want to make sure we're talking about
`the same things. When you say "melody," I'm
`thinking about the short piece that's submitted by
`the user.
`BY MR. DOVEL:
` Q Is it the case that what the Iwamura peak
`search does is to identify -- withdrawn.
` Well, in the reference database, what do
`you want to call those? Do you want to call those
`melodies, songs? What do you call those?
` A I would say it's -- it's audio works.
`It's music.
` Q Music?
` A Yeah. Yeah. So it's music works.
` Q Well, it's a set of music works.
` A That's right. So, for instance, there
`could be Schumann, there could be Beethoven, there
`could be others.
` Q The Iwamura database includes data about
`various musical works; right?
` A Yes.
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 216 of 384
`
`

`
`Page 217
`
`
`11
`2
`2
`3
`3
`4
`4
`5
`5
`6
`6
`7
`7
`8
`8
`9
`9
`10
`10
`11
`11
`12
`12
`13
`13
`14
`14
`15
`15
`16
`16
`17
`17
`18
`18
`19
`20
`21
`22
`23
`24
`25
`
`
` Q Before that search is run, each of thoseQ Before that search is run, each of those
`
`works is a possible match; right?
`works is a possible match; right?
` A Yes.
`A
`Yes.
` Q Would you agree that the Iwamura search,
`Q
`Would you agree that the Iwamura search,
`when it's run, it does a comparison of the unknown
`when it's run, it does a comparison of the unknown
`work to each of those possible matches?
`work to each of those possible matches?
`A
`To each of those possible music works,
` A To each of those possible music works,
`yes. An approximate comparison, just to be clear.
`yes. An approximate comparison, just to be clear.
` Q And by "approximate," you mean that it
`Q
`And by "approximate," you mean that it
`doesn't necessarily look at every bit of data in
`doesn't necessarily look at every bit of data in
`every musical work?
`every musical work?
` A It does not necessarily -- exactly -- use
`A
`It does not necessarily -- exactly -- use
`all the data, and then it uses only approximations
`all the data, and then it uses only approximations
`to the matching criterion.
`to the matching criterion.
` Q But it does examine each of the possible
`Q
`But it does examine each of the possible
`musical works -- or each of the musical works that
`musical works -- or each of the musical works that
`could be returned as a possible match?
`could be returned as a possible match?
` A Yes.
`A
`Yes.
` Q Would you agree that if we interpret
`"nonexhaustive" to mean a search that doesn't look
`at each of the possible matches, then this Iwamura
`search is an exhaustive search; it's not a
`nonexhaustive search?
` A Well, let me make sure we're on the same
`page.
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 217 of 384
`
`

`
`Page 218
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
` Okay. So when you talk about exhaustive
`versus nonexhaustive, it's not enough to say I'm
`comparing with every musical work in the record.
`You need to look at am I evaluating all the data
`within each musical work?
` Q When you did your analysis of
`exhaustive -- withdrawn.
` When you did your analysis in your
`declarations comparing the claims to the prior art,
`did you do it applying the definition of
`"exhaustive" that it means a comparison of all
`possible matches and all data within all possible
`matches?
` A So I had one construction that applied
`when I wrote my Declaration. And then following the
`Board's construction of "exhaustive" and
`"nonexhaustive search," I reevaluated all my
`opinions on this stand.
` Q That's not my question.
` Do you remember my question?
` A Yes. So everything I wrote on my report
`was under my construction of "exhaustive search" and
`"nonexhaustive search."
` Q I'm being more spec