Page 193
`
`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
`
`
`
`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
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
