`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`embodiment of Conwell t ha t 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 r obust hash
`
`value of our un known work?
`
`A
`
`Right .
`
`So just to give a specific
`
`example , you have an origina l song ; you apply this
`
`hashing - -
`
`this robust hashing algorithm ; you obtain
`
`a l28-bit sequence , which translates to Number 198 .
`
`Okay? So it goes in your database .
`
`Then you ha ve a version of that song which
`
`is not identica l. It could be , say , HP3 r ecorded at
`
`a different samp ling ra t e . You apply -- you go
`
`through this procedure ; you apply a robust hash
`
`algorithm , and i f indeed , you know , that song was
`
`16
`
`similar , with very high probability the hash values
`
`17 will be the same ; and therefore , we obtain , again ,
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`198 .
`
`So then we simply do a lookup in the
`
`table , and we have two songs that are simply a
`
`version of each other , a nd because they map to the
`
`exact same ident ifier , 198 , they are deemed to -- to
`
`be similar , and so they are in the neighborhood of
`
`each other .
`
`Q
`
`Element (c) says we ' re
`
`' - - - - - - - - - - - - - - - -
`TSG Re porting Worldwid
`PaS'-' 193 o f384
`
`NETWO RK- l EXH IBIT 2006
`Google Inc. v. Network-I Technologies, Lnc.
`IPR2015-OO345
`
`
`
`Page 193
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`embodiment o f Conwell t ha t does the neighbor search ,
`
`we ' ve got the Element (b) , those digitally created
`
`compact electronic representations of a first
`
`electronic work , because we ' re ta king a r obust hash
`
`value of our un known work?
`
`A
`
`Right .
`
`So just to give a specific
`
`example , you have an or iginal song ; you apply this
`
`hashing -- this robust hashing algorithm ; you obtain
`
`a l28-bit sequence , which translates to Number 198 .
`
`Okay? So it goes in your database .
`
`Then you ha ve a version of that song which
`
`is not identica l. It could be , say , HP3 r ecorded at
`
`a different samp ling ra t e . You apply -- you go
`
`through this procedure ; you apply a robust hash
`
`algorithm , and i f indeed , you know , that song was
`
`16
`
`similar , with very high probability the hash va lues
`
`17 will be the same ; and therefore , we obta in , again ,
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`198 .
`
`So then we simply do a lookup in the
`
`table , and we have two songs that are simply a
`
`version of each o ther , a nd because they map to the
`
`exact same ident ifier , 198 , they are deemed to -- to
`
`be similar , and so they are in the neighbo rhood o f
`
`each other .
`
`Q
`
`Element (c) says we ' re g oing to be
`
`TSG Re porting - Worldwide - 877 -7 02 - 9580
`PaS'-' 193 o f384
`
`
`
`Page 19 4
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`1
`
`8
`
`comparing the first electronic d a ta with the second
`
`digitally created compact electronic representation
`
`using a nonexhaustive neighbor sea r ch .
`
`Do you see that?
`
`Yes . That ' s 13(c) ; right?
`
`Yes .
`
`Yes .
`
`Is that , as you understand , Conwell - -
`
`A
`
`Q
`
`A
`
`Q
`
`9 withdrawn .
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`Is it your testimony that Conwell teaches
`
`Element (c) when it teaches using a robust hash
`
`approach , where it uses a lookup table to compare
`
`one hash value of the unknown work to hash values
`
`that are in the database?
`
`A
`
`Q
`
`Yes.
`
`In the embodiment in Conwell that --
`
`withdrawn .
`
`Would you agree that Conwell teaches some
`
`things are not a neighbor search?
`
`A
`
`1 will have to read the entire patent.
`
`I
`
`21 mean , clearly , the intent is to map similar songs to
`
`22
`
`23
`
`the same identifier .
`
`So that is the whole point of
`
`the patent .
`
`Q Well , if -- does Conwell teach using a
`
`25
`
`non robust hash?
`
`TSG Reporting - Worldwide - 877-702-9580
`P3ge 194 of384
`
`
`
`Page 195
`
`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 .
`
`1
`
`2
`
`]
`
`•
`
`5 Clearly , the int ent is to use -- it teaches ,
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`actually , using a robus t hash .
`
`It teaches that .
`
`Q
`
`A
`
`opinion
`
`Q
`
`A
`
`Q
`
`A
`
`Q
`
`right?
`
`A
`
`Q
`
`That ' s one o f the embodiments ; r i ght?
`
`It is the -- t he main one . That ' s my
`
`All right . The main one .
`
`Huh?
`
`It ' s the main one?
`
`Yes .
`
`But i t teaches other things as well ;
`
`It teaches many things , yes .
`
`Now , does it teach some things that would
`
`not be a neighbor search?
`
`A
`
`I wou l d 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
`
`Q
`
`Yes .
`
`All r i ght . Then let me see how that maps
`
`TSG Reporting - Worldwide - 877-702-9580
`PaS'-' 19S ofJ84
`
`
`
`Page 196
`
`1
`
`2
`
`]
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`"
`
`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
`
`Q
`
`Yes .
`
`Among those are " First electronic data ,
`
`including a first digitally created compact
`
`electronic representation of one or more referenced
`
`electronic works "; righ t ?
`
`A Yes .
`
`Q
`
`In Conwell in the embodiment you ' re
`
`pointing to , the " electronic works " are various
`
`digital songs , right , in a database?
`
`A
`
`Q
`
`Yes .
`
`In Conwell in the embodiment you ' re
`
`pointing to , is the " compact electronic
`
`representation " the hash value?
`
`A
`
`Q
`
`Yes , it is .
`
`Now , the next element ,
`
`(a) (2) , is
`
`"electronic data related to an action. "
`
`Do you see that?
`
`Yes .
`
`And it relates to an advertisement .
`
`A
`
`Q
`
`Do you see that?
`
`A
`
`Yes .
`
`TSG Reporting - Worldwide - 877-702-9580
`PaS'-' 196 ofJ84
`
`
`
`Page 197
`
`1
`
`2
`
`]
`
`•
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`Q
`
`Then Element (h) says , "Obtaining a second
`
`digitally created compact electronic representation
`
`of a first elect ronic work ."
`
`A
`
`Q
`
`Yes .
`
`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 o b tain 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 :
`o Let ' s continue with ~lement (c) of
`It carries over to
`Claim 13 . which is on page 50 .
`
`TSG Reporting - Worldwide - 877-702-9580
`Pag<.' 197 ofJ84
`
`
`
`Page 198
`
`1
`
`2
`
`]
`
`4
`
`5
`
`page 51 of your Declara t ion .
`
`In the embodiment in Conwell that you
`
`assert discloses the neighbor search -- wi t hdrawn .
`
`The c l aim requires in Element (c) that
`
`we ' re going to compare the first electronic data
`
`6 with the second digitally created compact electronic
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`representation using a nonexhaustive neighbor
`
`search .
`
`A
`
`Q
`
`That ' s part of it ; right?
`
`Yes .
`
`In your -- in the embodiment in Conwell
`
`that you assert discloses this claim limitation,
`
`1] what is it that constitutes the second digitally
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`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
`
`Q
`
`It is comparing the hash values .
`
`Then , when it says Musing 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
`PaS'-' 198 ofJ84
`
`
`
`Page 199
`
`1
`
`Q
`
`Is the search that ' s performed --
`
`2 withdrawn.
`
`3
`
`4
`
`5
`
`6
`
`1
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`11
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`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 t he item canno t be
`
`found .
`
`Q When you say it ' s a search , wh y isn ' t --
`
`isn ' t a search something where you ' ve got to loo k
`
`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 tabie constitutes
`
`a search?
`
`A
`
`You can do a search from a lookup table.
`
`It's very easy.
`
`Q
`
`Is it the case that -- withdrawn .
`
`Is it your assertion that the
`
`nonexhaustive neighbor search here is the search
`
`that ' S done when we use the hash value lookup table?
`
`TSG Reporting - Worldwide - 811-102-9580
`P3ge ] 9'9 oDS4
`
`
`
`Page 200
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`A
`
`Yes. So if you have computed -- say our
`
`identifier was 198 in decimal representation, we
`
`simply have to look up, is that number in t he left
`
`column of the table . And that ' s a straightforward
`
`thing to do ; so the search is trivial .
`
`Q
`
`In the -- is that t he only search that ' s
`
`disclosed in Conwell , this loo kup table?
`
`A
`
`I would have t o read the whole patent .
`
`The main poi nt is what I just said here .
`
`Q
`
`Is it the case that -- tha t -- withdrawn.
`
`If we look at the search you ' ve identified
`
`as constituting the neighbor search , it's a search
`
`13 wherein we start with a hash value and then go to a
`
`14
`
`15
`
`16
`
`17
`
`19
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`lookup table to see if it appears in that table; is
`
`that right?
`
`A
`
`Q
`
`Yes.
`
`Is the result of that going to be either
`
`an exact match of the hash or a determination that
`
`it does not exist?
`
`A
`
`Q
`
`Yeah .
`
`It's either the table or it ' s not.
`
`Are we ever going to have a circumstance
`
`where we look in the table for that hash value and
`
`we conclude that the hash value does not appear , but
`
`here ' s one that ' s pretty close, and we ' ll return
`
`that one?
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 200 o f 384
`
`
`
`Page 201
`
`1
`
`2
`
`3
`
`.,
`
`5
`
`6
`
`1
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`A
`
`It ' s not -- it does not -- Conwell does
`
`not disclose that .
`
`Q
`
`It we return the -- withdrawn .
`
`So the only search that ' s disclosed in
`
`in Conwell using the hash values is one where we
`
`look for an exact match , and it ' s either there or it
`
`isn ' t ; is that right?
`
`A Well, an exact match of that identifier,
`
`Like 19B , is it there or not?
`
`If I have 199, I
`
`would say it ' s not there .
`
`Q
`
`Is it the case that the Conwell search
`
`looks for an exact match of t he digi t ally created
`
`compact electronic representation with one of the
`
`items that constitutes the first electronic data?
`
`A
`
`So that ' s understood , when we say "exact
`
`16 match ," we ' re talking about the exact match of the
`
`17
`
`identifier. Okay?
`
`It ' s not an exact match of the
`
`18 works.
`
`19
`
`20
`
`21
`
`22
`
`23
`
`" 2S
`
`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
`
`Q
`
`A
`
`Yes.
`
`Now let ' s take a look at Claim I .
`
`In the ' 179 patent?
`
`TSG Reporting - Worldwide - 877-702-9580
`P3ge 20 I o DS4
`
`
`
`Page 202
`
`1
`
`2
`
`]
`
`Q
`
`Yes .
`
`In Claim 1 , we have a database
`
`comprising first electronic data related to
`
`identifica t ion of one o r more referenced elect r onic
`
`4 works .
`
`5
`
`6
`
`7
`
`8
`
`I n the Conwel l embodiment that you
`
`identify as disclosing t his claim , what is it that
`
`constitutes the f irst e le ctronic data related to
`
`identification of one or more referenced electronics
`
`9 works?
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`A Well , it is i l lustrated by the table . So
`
`it is the identif iers that we ~ - that are
`
`represented in the l eft col umn of Figure 3 . These
`
`are the identif i ers derived from content .
`
`Q
`
`Is the -- are the identifiers in the
`
`embodiment you say anticipates t h is claim r obust
`
`hash values?
`
`A
`
`That p art has nothing to do with
`
`robustness .
`
`So it anticipates it because it
`
`describes identifiers .
`
`Q Well , you understand that for the cla im as
`
`a whole , to anticipate it , you ' ve got to use the
`
`same thing all t he 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 Re p orting - worldwide - 877 - 702-9580
`Pa~ 202 of 384
`
`
`
`Page 203
`
`1
`
`2
`
`]
`
`•
`
`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 underst.and .
`
`Q All ri ght .
`
`Is it the case that in the
`
`identify you identify that anticipates in Conwell ,
`
`t.he first electronic data related to identification
`
`of one o r more referenced electronic works is the
`
`database that includes t he 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 comput er system extracted feat.ures
`
`of a first electronic work. "
`
`In the embodiment that you say anticipates
`
`the claim , the embodiment of Conwell , what is it
`
`that constitutes the e xtracted features of a first
`
`electronic work?
`
`A
`
`Q
`
`A
`
`Q
`
`That ' s obtained by the hashing algorithm .
`
`It's going to be the hash values?
`
`Yes .
`
`So if we have an unknown work , in order to
`
`create -- to e x tract 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
`Pa~ 203 of384
`
`
`
`Page 204
`
`1
`
`2
`
`3
`
`Q
`
`Then in Element
`
`(c) , it requires
`
`identifying the first electronic work by comparing
`
`the extracted features of the first electronic work
`
`4 with the first electronic data in the database using
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`a nonexhaustive neighbor search .
`
`Do you see that?
`
`A Right .
`
`Q What are the t wo 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 f ormat , as illustrated ~n
`
`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 tha t
`
`right?
`
`A
`
`Yes . Yes .
`
`TSG Reporting - Worldwide - 877-702-9580
`Pa~ 204 ofJ84
`
`
`
`Page 205
`
`1
`
`2
`
`]
`
`•
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`Q
`
`And i t may be t hat we ' re comparing 128-bit
`
`representations or decimal representations?
`
`A
`
`Q
`
`It ' s immaterial .
`
`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 da t a in the database?
`
`A
`
`Q
`
`Yes .
`
`Then t he next part is using a
`
`nonexhaustive neighbor search.
`
`Is the -- using a nonexhaustive neighbor
`
`search , is that the
`
`using the lookup table to see
`
`14 whether the hash value of the unknown work appears?
`
`15
`
`16
`
`17
`
`18
`
`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
`
`19 marked as Exhibit 1012. This is the Iwamura patent .
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`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 I wamura , it teaches a system that
`
`TSG Reporting - Worldwide - 877-702-9580
`Pa~ 20S of 384
`
`
`
`Page 206
`
`1
`
`2
`
`]
`
`•
`
`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 r ight?
`
`A
`
`Q
`
`Yes .
`
`If we l ook at t he abstract of I wamura , it
`
`says , " A remote music database with melody
`
`information is searched "; is that right?
`
`A
`
`Q
`
`Yes .
`
`In doing that searches -- in doing that
`
`search , what we ' r e -- a ga in , we ' re starting with a
`
`query or an unknown melody ; is t hat right?
`
`A Yes .
`
`Q When we do the search t ha t ' s desc r ibed in
`
`Iwamura , is the -- are
`
`t he -- the possib le matches
`
`the s e t of the melodies in the database in I wamu ra ?
`
`A
`
`Q
`
`Yeah , that you try to match melodies .
`
`Now , the search that ' s - -
`
`that ' s described
`
`in Iwamu r a -- i n fact ,
`
`I ' ll have you loo k at
`
`Column 7 .
`
`Now ,
`
`Iwamura t eaches a type of peak note
`
`search o r what i t calls a peak search ; is t hat
`
`right?
`
`A Among other thi ng s , ye s .
`
`Q Well ,
`
`i n addi ti on to t e aching a peak
`
`search , it teaches varia t ions of that where you
`
`would use dips rather than peaks . That would be one
`
`TSG Re porting - Worldwide - 877 - 702-9580
`Pa~ 206 of 384
`
`
`
`Page 207
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`"
`
`25
`
`variation of it?
`
`A
`
`Q
`
`Or both .
`
`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 th inks is -- what Iwamura
`
`describes as a -- as a new app roach or a novel
`
`search that involves searching with peaks or dips
`
`and so on ; right?
`
`A
`
`Q
`
`Right .
`
`Now, in the type of search that'S
`
`identified and described in Iwamura, what it does is
`
`it does a comparison of the unknown melody to each
`
`of the melody patterns that are in the melody
`
`database; right?
`
`A
`
`Q
`
`Yes .
`
`For example , if we look at -- what that
`
`means , first , is that we are going to ta ke our
`
`TSG Reporting - Worldwide - 877-702-9580
`P3ge 207 of 384
`
`
`
`Page 208
`
`1
`
`2
`
`]
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`unknown melody , and we ' r e going to use this peak
`
`search comparison method to see whether it matches
`
`e a ch of the melod ies that are - -
`
`that are in the
`
`database ; right?
`
`A
`
`Q
`
`Yes .
`
`Would you agree that it -- that t he search
`
`that ' s identified -- wi thdrawn .
`
`Is it the case that the search identified
`
`in Iwamura does a compa rison of all of the possib le
`
`10 matches?
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`A
`
`NO .
`
`Q Why not?
`
`A Well , because t hey define what a match is .
`
`Let me retrieve that .
`
`Yes .
`
`I f 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 f or
`
`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 computa t ion
`
`time .
`
`So it ' s an approximate search .
`
`TSG Reporting - Worldwide - 877 - 702-9580
`Pa~ 208 of 384
`
`
`
`Page 209
`
`1
`
`2
`
`]
`
`4
`
`5
`
`Q All right . Would you agree that they loo k
`
`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
`
`I s it the case that the search proceeds --
`
`6 withdrawn .
`
`7
`
`The
`
`in Iwamura , is it the case that
`
`8 what it identifies as a match would be one of the
`
`9 melodies in the database?
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`A Yes .
`
`Q
`
`Is the set o f all possible matches in the
`
`Iwamuta search the set o f a ll melodies in the
`
`database?
`
`A
`
`No . Because it does not evaluate , for
`
`instance , every possible shift . And whenever it
`
`evaluates a shi f t , it does not use the matching
`
`criter i on that they ideally would like to use for an
`
`e xact match or an e xa c t search ; they use an
`
`approximation to it .
`
`Q
`
`Is it the case t hat if we
`
`withdrawn.
`
`If a melody is identified as a match using
`
`the Iwamura algorithm, would the melody then be --
`
`23 withdrawn .
`
`24
`
`25
`
`If the lwamura search approach determines
`
`that a particular melody meets the matching
`
`TSG Re p orting - Worldwide - 877 - 702-9580
`Pa~ 209 of 384
`
`
`
`Page 210
`
`I
`
`criterion , would i t
`
`t hen identify the melody as a
`
`2 match?
`
`]
`
`•
`
`5
`
`6
`
`7
`
`A
`
`Q
`
`It would declare it as a match .
`
`And when it declares it as a match , what
`
`does it do? How does it do that?
`
`A Well , it evaluates the matching c r iterion
`
`and then it -- among al l the possible potential
`
`8 matches it evaluates ,
`
`i t tries to find the best fit ;
`
`right?
`
`9
`
`10
`
`Let ' s say it wants to find one single
`
`11 matching melody , o kay, then it would try to find the
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`one that minimi zes that -- the value of t hat
`
`cri te rion .
`
`Q
`
`It ' s going to f ind the one that either --
`
`it ' s go ing to identify t he melodies that are either
`
`exact matches , because the absolute total difference
`
`is zero , or the melodies that have the least
`
`absolute dif f e r ence ; is t hat right?
`
`A
`
`That ' s right . The -- the perfect match is
`
`a ra re , ideal c ase .
`
`So usually , the c riteri on will
`
`not have zero va l ue , and it tries to fin d the best
`
`ma tch .
`
`Q
`
`I s it the case that in Iwamura ,
`
`t hat the
`
`24 match that ' s ret urned is an identification of a
`
`25
`
`particular melod y in the re f erence database?
`
`TSG Reporting - Worldwide - 877 - 702 - 9580
`PaS'-' 21 0 o fJ84
`
`
`
`Page 211
`
`1
`
`2
`
`]
`
`4
`
`A
`
`Yes .
`
`It iden t i f ies 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
`
`5 matches the melody , or does it declare that the song
`
`6 matches or is the best match for the melody?
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`A Well ,
`
`i t searches for the song t i tle ;
`
`right? So the melody that ' s input by the user , you
`
`know , you try to find a match somewhere in -- in t he
`
`song .
`
`It does not really matter where . The
`
`important thing is to identify the song .
`
`Now , what the -- if we look at Claim 25 o f
`
`patent - - get t hat language in front of
`
`Claim 25 of the patent . Oh, it ' s here .
`
`I ' ll find it for you in your report .
`
`The patent is here unless you want me to
`
`look 1n my char t .
`
`MR . DOVEL : The patent is fine .
`
`If you look a t Claim 25 , if you can -- i f
`
`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
`
`T5G Reporting - Worldwide - 877-702-9580
`PaS'-' 21 1 ofJ84
`
`
`
`Page 212
`
`1
`
`2
`
`]
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`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 f ind the right page for you here .
`
`Just one second .
`
`Let ' s go take a look at page 75 .
`
`10 Actually,
`
`I ' ll have you look at the claim chart .
`
`So
`
`11
`
`12
`
`1]
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`let ' s go over to page 80 .
`
`Do you see there ' s claim language for
`
`Claim 15 that ' s reproduced there?
`
`A
`
`Q
`
`A
`
`Q
`
`On page aO?
`
`Yes .
`
`Uh-huh .
`
`In Element 15 -- sorry . Claim -- in
`
`Claim 15 , Element (b) requires doing a nonexhaustive
`
`search identifying a neighbor.
`
`Do you see that?
`
`Yes .
`
`And you incorporate by reference from
`
`A
`
`Q
`
`Claim Element l(c) .
`
`Do you see that?
`
`A
`
`Yes .
`
`TSG Reporting - Worldwide - 877-702-9580
`PaS'-' 212 ofJ84
`
`
`
`Page 213
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`l6
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`Q
`
`And if we go back to p age 78 and 79 ,
`
`you ' ve got your claim chart that identifies what you
`
`point to there as identifying as the nonexhaustive
`
`search; right?
`
`A
`
`Q
`
`correct .
`
`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 ' 5 back on page 27 .
`
`Now , in paragraph 7 1, you pre sent some
`
`analysis to demonstrate how I wamura teaches a search
`
`that can be nonexhaustive .
`
`Do you see that?
`
`Yes .
`
`Here , what you identify as a nonexhaustive
`
`A
`
`Q
`
`search is searching that uses peak notes .
`
`A
`
`Q
`
`" Here , " you mean on page 27?
`
`Yes.
`
`In paragraph 71 . We ' re focused on
`
`that first .
`
`A
`
`Q
`
`I give that example , yes .
`
`Would you agree that in Iwamura ' s peak
`
`search note, that it does search across each melody
`
`in the melody database?
`
`TSG Reporting - Worldwide - 877-702-9580
`P3ge 2 13 of 384
`
`
`
`Page 214
`
`1
`
`A
`
`Yes, it does.
`
`It ' s going to compare it
`
`2 with every song in the database , typically .
`
`3
`
`4
`
`5
`
`6
`
`1
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`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 pea k note search
`
`14 will do a comparison to each of the possible
`
`15 melodies that could be a match?
`
`l6
`
`11
`
`18
`
`A
`
`You give only one part of -- of the -- the
`
`question .
`
`It does not evaluate all the possible
`
`19 melodies because it does not use all the data . As
`
`20
`
`21
`
`22
`
`23
`
`" 25
`
`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
`
`Q
`
`Yes .
`
`And when the Iwamura peak note search
`
`TSG Reporting - Worldwide - 811-102-9580
`P3ge 2 14 of384
`
`
`
`Page 215
`
`1
`
`2
`
`]
`
`•
`
`5
`
`performs -- before it ' s run, we ' ve got an unknown
`
`work in front of us ; right?
`
`A
`
`Q
`
`Uh-huh .
`
`And at that point, the possible melodies
`
`are the melodies that are -- rather , the possible
`
`6 matches are all the melodies that are in the
`
`7
`
`8
`
`database ; right?
`
`A
`
`I just want to make clear - - okay . A
`
`9 melody in the context of Iwamura is a short segment ;
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`right? And then you have a whole audio piece which
`
`could be much longer . And so you ' re comparing tha t
`
`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
`
`21 melodies that are in the database ; right?
`
`22
`
`23
`
`24
`
`25
`
`A
`
`I ' m making
`
`MR . ELACQUA : Objection .
`
`'l'HE WITNESS :
`
`1 ' m making a distinction ,
`
`jus t
`
`like Iwamura does , between the melody itself and the
`
`TSG Reporting - Worldwide - 877-702-9580
`PaS'-' 215 o fJ84
`
`
`
`Page 21 6
`
`1
`
`2
`
`]
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`1]
`
`entire piece o f music .
`
`So the melody that ' s input
`
`by the user cou l d be very short , say , a few seconds ,
`
`five seconds . The piece of music in the da tabase
`
`could be much longer .
`
`So I want to ma ke sure we ' re tal king about
`
`the same things . When you say "melody ," I ' m
`
`thinking about t he short piece that ' s submi tted by
`
`the user .
`
`BY MR . DOVEL :
`
`Q
`
`Is it the case t hat what the Iwamura pea k
`
`search does is to identi f y ~~ wi t hdrawn .
`
`Wel l ,
`
`i n the re f erence database , what do
`
`you want to call t hose? Do you want to call those
`
`14 melodies , songs? Wha t do you call those?
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`A
`
`I would say i t ' s ~~ it ' s audio wo r ks .
`
`It ' s mu sic .
`
`Q
`
`A
`
`Q
`
`A
`
`Mus i c?
`
`Yeah . Yeah .
`
`So it ' s music wor ks .
`
`Well , it ' s a set of music works .
`
`That ' s righ t. So , fo r instance , there
`
`could be Schumann , there could be Beethoven , there
`
`could be others .
`
`Q
`
`The Iwamura database includes data about
`
`various musical works ; r ight?
`
`A
`
`Yes .
`
`TSG Reporting ~ Worldwide ~ 877 ~ 702~9S80
`PaS'-' 21 6 ofJ84
`
`
`
`Page 217
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`Q
`
`Before that search is run , each of those
`
`works is a possible match ; right?
`
`A
`
`Yes .
`
`Q Would you agree that the Iwamura search ,
`
`when it ' s run, it does a comparison of the unknown
`
`work to each of t hose possible matches?
`
`A
`
`To each of those possible music works ,
`
`yes . An approximate comparison , just to be clear .
`
`Q
`
`And by " approximate ," you mean that it
`
`doesn ' t necessarily look at every bit of data in
`
`every musical work?
`
`A
`
`It does not necessarily -- exactly -- use
`
`all the data , and then it uses only approximations
`
`to the matching criterion .
`
`Q
`
`But it does examine each of the possible
`
`16 musical works
`
`or each of t he musical works that
`
`17
`
`19
`
`19
`
`20
`
`21
`
`22
`
`23
`
`could be returned as a possible match?
`
`A
`
`Q
`
`Yes .
`
`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
`
`nonexhaus t ive search?
`
`A
`
`Well ,
`
`l et me make sure we ' re on the same
`
`25
`
`page.
`
`TSG Reporting - Worldwide - 877-702-9580
`Page 2 17 of384
`
`
`
`Page 218
`
`1
`
`2
`
`]
`
`4
`
`Okay .
`
`So when you tal k about e x haus t ive
`
`versus nonexhaustive ,
`
`i t ' s not enough to say I ' m
`
`comparing with every musical wor k in the record .
`
`You need to loo k at am I evaluating all the data
`
`5 within each musical work?
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`Q When you did your analysis of
`
`e xhaustive -- withdrawn .
`
`When you did your analysis in your
`
`declarations comparing the claims to the prior ar t ,
`
`did you do it a pp lying the de f inition of
`
`"e xhaustive " that it means a comparison of all
`
`possible matches and al l data within all possible
`
`13 ma tches?
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`"
`
`25
`
`A
`
`So I had one construction that applied
`
`when I wrote my Declara ti on . And then following the
`
`Board I 5 construction of "e x haustive " and
`
`" none xhaustive 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
`
`wa s under my constru