Ex Parte Lee et al

Patent Trial and Appeal Board

Aug 29, 2018

14383341 (P.T.A.B. Aug. 29, 2018)

UNITED STA TES p A TENT AND TRADEMARK OFFICE
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR
14/383,341 09/05/2014
21186 7590 09/06/2018
SCHWEGMAN LUNDBERG & WOESSNER, P.A.
P.O. BOX 2938
MINNEAPOLIS, MN 55402
UNITED ST A TES OF AMERICA
Siu WaLee
UNITED STATES DEPARTMENT OF COMMERCE
United States Patent and Trademark Office
Address: COMMISSIONER FOR PATENTS
P.O. Box 1450
Alexandria, Virginia 22313-1450
www .uspto.gov
ATTORNEY DOCKET NO. CONFIRMATION NO.
3145.067US1 3561
EXAMINER
KIM, JONATHAN C
ART UNIT PAPER NUMBER
2659
NOTIFICATION DATE DELIVERY MODE
09/06/2018 ELECTRONIC
Please find below and/or attached an Office communication concerning this application or proceeding.
The time period for reply, if any, is set in the attached communication.
Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the
following e-mail address(es):
uspto@slwip.com
SLW@blackhillsip.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte SIU WA LEE, LING CEN, HAIZHOU LI,
Y AOZHU PAUL CHAN, and MINGHUI DONG
Appeal2018-000374
Application 14/383,341 1
Technology Center 2600
Before LARRY J. HUME, JAMES W. DEJMEK, and
JOHN D. HAMANN, Administrative Patent Judges.
DEJMEK, Administrative Patent Judge.
DECISION ON APPEAL
Appellants appeal under 35 U.S.C. § 134(a) from a Final Rejection of
claims 1-19. We have jurisdiction over the pending claims under 35 U.S.C.
§ 6(b).
We affirm.
1 Appellants identify the Agency for Science, Technology and Research as
the real party in interest. Br. 3.
Appeal2018-000374
Application 14/383,341
STATEMENT OF THE CASE
Introduction
Appellants' disclosed and claimed invention generally relates to
"template-based personalized singing synthesis." Spec. ,r 2. In particular,
the disclosed approach uses speech-to-singing (STS) synthesis. Spec. ,r 4.
According to the Specification, in STS synthesis, the synthesized singing of
a speaker preserves the timbre of the speaker such that the synthesized
singing will sound like it is being sung by the speaker. Spec. ,r 6. In a
disclosed embodiment, characteristics of a singing voice for a first individual
(e.g., a professional singer) are derived and the vocal characteristics of a
voice of a second individual (i.e., a speaker/unprofessional singer) are
modified in response to the characteristics of the first individual's singing
voice to generate a synthesized singing voice for the second individual.
Spec. ,r 9.
Claim 1 is representative of the subject matter on appeal and is
reproduced below with the disputed limitation emphasized in italics:
1. A method for speech-to-singing synthesis comprising:
deriving characteristics of a singing voice for a first
individual; and
modifying vocal characteristics of a voice for a second
individual in response to the characteristics of the singing voice
of the first individual to generate a synthesized singing voice for
the second individual,
wherein modifying the vocal characteristics of the voice of
the second individual comprises aligning the voice of the second
individual to match the singing voice of the first individual.
2
Appeal2018-000374
Application 14/383,341
The Examiner's Rejections
1. Claims 1-19 stand rejected under 35 U.S.C. § 101 as being
directed to patent-ineligible subject matter. Final Act. 8.
2. Claims 1, 3-6, and 8 stand rejected under 35 U.S.C. § I02(a) as
being anticipated by Kawashima et al. (US 6,836,761 Bl; Dec. 28, 2004)
("Kawashima"). Final Act. 9-12.
3. Claim 19 stands rejected under 35 U.S.C. § I02(a) as being
anticipated by Haupt et al. (US 2013/0019738 Al; Jan. 24, 2013 (filed
July 22, 2011)) ("Haupt"). Final Act. 12-13.
4. Claim 2 stands rejected under 35 U.S.C. § I03(a) as
unpatentable over Kawashima and Haupt. Final Act. 13-14.
5. Claim 7 stands rejected under 35 U.S.C. § I03(a) as
unpatentable over Kawashima and Oytun Tiirk and Marc Schroder,
Evaluation of Expressive Speech Synthesis With Voice Conversion and
Copy Resynthesis Techniques, IEEE Transactions on Audio, Speech, and
Language Processing, Vol. 18, No. 5, 965-73 (July 2010) ("Tiirk"). Final
Act. 14--15.
6. Claims 9, 10, and 14--17 stand rejected under 35 U.S.C.
§ I03(a) as unpatentable over Haupt and Ling Cen et al., Segmentation of
Speech Signals in Template-based Speech to Singing Conversion, APSIPA
ASC 2011 Xi'an (2011) ("Cen"). Final Act. 15-18.
7. Claims 11-13 stand rejected under 35 U.S.C. § I03(a) as
unpatentable over Haupt, Cen, and Tiirk. Final Act. 19-21.
8. Claim 18 stands rejected under 35 U.S.C. § I03(a) as
unpatentable over Haupt, Cen, and Nakano et al. (US 2009/0306987 Al;
Dec. 10, 2009) ("Nakano"). Final Act. 21-22.
3
Appeal2018-000374
Application 14/383,341
ANALYSIS 2
Rejection under 35 U.S.C. § 1 OJ
Appellants dispute the Examiner's conclusion that the pending claims
are directed to patent-ineligible subject matter. Br. 7-9. In particular,
Appellants argue the claims involve converting digital signals from one form
(i.e., speech) to another (i.e., synthesized singing) to achieve a technical
result-"to create a professional-quality singing voice for poor singers using
the timbre of the speaker's voice while applying the melody of a
professional voice." Br. 8. Appellants assert the claims provide a major
advancement in speech to singing technology. Br. 8. Additionally,
Appellants analogize the pending claims to the claims at issue in McRO, Inc.
v. Bandai Namco Games America Inc., 837 F.3d 1299 (Fed. Cir. 2016).
Br. 9. More specifically, Appellants argue the court in McRO concluded the
claims in that case were not directed to an abstract idea but rather recited
"'perform[ing] a distinct process to automate a task previously performed by
humans' and 'us[ing] a combined order of specific rules that renders
information into a specific format that is then used and applied to create
desired results."' Br. 9. Appellants contend the pending claims similarly
apply rules for speech-to-singing synthesis. Br. 9.
The Supreme Court's two-step framework guides our analysis of
patent eligibility under 35 U.S.C. § 101. Alice Corp. Pty. v. CLS Bank Int'!,
134 S. Ct. 2347, 2355 (2014). If a claim falls within one of the statutory
categories of patent eligibility (i.e., a process, machine, manufacture, or
2 Throughout this Decision, we have considered the Appeal Brief, filed
June 5, 2017 ("Br."); the Examiner's Answer, mailed August 25, 2017
("Ans."); and the Final Office Action, mailed May 6, 2016 ("Final Act."),
from which this Appeal is taken. Appellants did not file a Reply Brief.
4
Appeal2018-000374
Application 14/383,341
composition of matter) then the first inquiry is whether the claim is directed
to one of the judicially recognized exceptions (i.e., a law of nature, a natural
phenomenon, or an abstract idea). Alice, 134 S. Ct. at 2355. If so, the
second step is to determine whether any element, or combination of
elements, amounts to significantly more than the judicial exception. Alice,
134 S. Ct. at 2355.
Although the independent claims each broadly fall within the statutory
categories of patentability, the Examiner concludes the claims are directed to
a judicially recognized exception-i.e., an abstract idea. Final Act. 8. In
particular, the Examiner concludes the claims are directed to "modifying
vocal characteristics of a voice for a second individual." Ans. 2; see also
Final Act. 8. The Examiner explains such modification consists of a series
of mathematical operations. Final Act. 8. Further, the Examiner finds the
claims only require generic computer elements performing "generic
computer functions that are well-understood, routine and conventional
activities previously known to the industry" and concludes, therefore, the
claims do not recite significantly more to transform the claims into patent-
eligible subject matter. Final Act. 8.
Instead of using a definition of an abstract idea, "the decisional
mechanism courts now apply is to examine earlier cases in which a similar
or parallel descriptive nature can be seen-what prior cases were about, and
which way they were decided." Amdocs (Isr.) Ltd. v. Openet Telecom, Inc.,
841 F.3d 1288, 1294 (Fed. Cir. 2016) (citing Elec. Power Grp., LLC v.
Alstom S.A., 830 F.3d 1350, 1353-54 (Fed. Cir. 2016)); accord United
States Patent and Trademark Office, July 2015 Update: Subject Matter
Eligibility 3 (July 30, 2015), https://www.uspto.gov/sites/default/files/
5
Appeal2018-000374
Application 14/383,341
documents/ieg-july-2015-update.pdf (instructing Examiners that "a claimed
concept is not identified as an abstract idea unless it is similar to at least one
concept that the courts have identified as an abstract idea."). As part of this
inquiry, we must "look at the 'focus of the claimed advance over the prior
art' to determine if the claim's 'character as a whole' is directed to excluded
subject matter." Affinity Labs of Tex., LLC v. DirecTV, LLC, 838 F.3d 1253,
1257 (Fed. Cir. 2016).
Our reviewing court has concluded that abstract ideas include the
concepts such as the collection and analysis of information. Elec. Power,
830 F.3d at 1353; see also Elec. Power, 830 F.3d at 1354 ("analyzing
information by steps people go through in their minds, or by mathematical
algorithms, without more, as essentially mental processes within the
abstract-idea category"). Similarly, "collecting, displaying, and
manipulating data" is an abstract idea. Intellectual Ventures I LLC v.
Capital One Fin. Corp., 850 F.3d 1332, 1340 (Fed. Cir. 2017) (additionally
concluding that "customizing information and presenting it to users based on
particular characteristics" is an abstract idea ( citing Intellectual Ventures I
LLC v. Capital One Bank (USA), 792 F.3d 1363, 1370 (Fed. Cir. 2015)).
Further, our reviewing court has concluded that a process that employs
mathematical algorithms to manipulate existing information to generate
additional information is not patent eligible. Digitech Image Techs., LLC v.
Elec.for Imaging, Inc., 758 F.3d 1344, 1351 (Fed. Cir. 2014); see also
Parker v. Flook, 437 U.S. 584, 595 (1978) ("If a claim is directed essentially
to a method of calculating, using a mathematical formula, even if the
solution is for a specific purpose, the claimed method is [patent ineligible].")
(internal quotations omitted).
6
Appeal2018-000374
Application 14/383,341
Further, merely combining several abstract ideas does not render the
combination any less abstract. RecogniCorp, LLC v. Nintendo Co., 855 F.3d
1322, 1327 (Fed. Cir. 2017) ("Adding one abstract idea (math) to another
abstract idea ... does not render the claim non-abstract."); see also
FairWarning IP, LLC v. Iatric Sys., Inc., 839 F.3d 1089, 1093-94 (Fed. Cir.
2016) ( determining the pending claims were directed to a combination of
abstract ideas).
Here, Appellants' claims are directed to generating a synthesized
singing voice by modifying vocal characteristics of one individual based on
the characteristics derived from the singing voice of another individual.
Deriving characteristics of a signal ( e.g., a singing voice) is similar to the
abstract idea of analyzing data. See Elec. Power, 830 F.3d at 1353-54;
Intellectual Ventures, 850 F.3d at 1340. Similarly, modifying a signal ( e.g.,
the vocal characteristics of another's voice) based on the results of analyzing
a first signal is also similar to the abstract idea of manipulating data using a
mathematical relationship. See Intellectual Ventures, 850 F.3d at 1340;
Digitech, 758 F.3d at 1351. Neither these steps, nor those recited in the
dependent claims alters the character of the claims, which is directed to an
abstract idea.
To the extent Appellants assert the pending claims are not directed to
an abstract idea, but are similar to those held patent eligible in McRO (see
Br. 9), we disagree. As our reviewing court noted in McRO, it is important
to determine "whether the claims ... focus on a specific means or method
that improves the relevant technology or are instead directed to a result or
effect that itself is the abstract idea and merely invoke generic processes and
machinery." 837 F.3d at 1314. In other words, a claim that defines the way
7
Appeal2018-000374
Application 14/383,341
by which a computer-related result is achieved is distinguishable from a
patent-ineligible claim that simply describes a result. Here, we agree with
the Examiner that the claims recite at a high level deriving characteristics
and modifying vocal characteristics in response to the derived
characteristics, but do not otherwise recite how such derivation or
modification is done. Ans. 5---6. Accordingly, we are unpersuaded of
Examiner error.
Because we determine the claims are directed to an abstract idea or
combination of abstract ideas, we analyze the claims under step two of Alice
to determine if there are additional limitations that individually, or as an
ordered combination, ensure the claims amount to "significantly more" than
the abstract idea. Alice, 134 S. Ct. at 2355 (citing Mayo Collaborative
Servs. v. Prometheus Labs., Inc., 566 U.S. 66, 72, 78-79 (2012)). The
implementation of the abstract idea involved must be "more than [the]
performance of 'well-understood, routine, [and] conventional activities
previously known to the industry.'" Content Extraction and Transmission
LLC v. Wells Fargo Bank N.A., 776 F.3d 1343, 1347--48 (Fed. Cir. 2014)
(quoting Alice, 134 S. Ct. at 2359) (alteration in original). "Whether
something is well-understood, routine, and conventional to a skilled artisan
at the time of the patent is a factual determination." Berkheimer v. HP Inc.,
881 F.3d 1360, 1369 (Fed. Cir. 2018).
Appellants do not dispute the Examiner's findings that the claims
recite "generic computer functions that are well-understood, routine and
conventional activities." See Final Act. 8. Appellants do not advance any
arguments related to the Examiner's findings and conclusions related to
whether the claims recite significantly more to transform the abstract idea
8
Appeal2018-000374
Application 14/383,341
into a patent-eligible application. Accordingly, we are not apprised of
Examiner error.
For the reasons discussed supra, we sustain the Examiner's rejection
under 35 U.S.C. § 101 of independent claim 1. For similar reasons, we also
sustain the Examiner's rejection under 35 U.S.C. § 101 of independent
claims 9 and 19, which recite similar limitations and were not argued
separately. See Br. 9; see also 37 C.F.R. § 4I.37(c)(l)(iv) (2017).
Additionally, we sustain the Examiner's rejection under 35 U.S.C. § 101 of
claims 2-8 and 10-18, which depend directly or indirectly therefrom and
were not argued separately. See 37 C.F.R. § 4I.37(c)(l)(iv).
Rejections under 35 U.S.C. § 102
a. Kawashima ( claims 1, 3---6, and 8)
Appellants contend Kawashima fails to disclose "modifying the vocal
characteristics of the voice of the second individual comprises aligning the
voice of the second individual to match the singing voice of the first
individual," as recited in claim 1. Br. 9-11. Instead, Appellants argue
Kawashima discloses allocating vocal characteristics of a first individual to
the vocal characteristics of a second individual, but does not align the voice
of the second individual to match the singing voice of the first individual.
Br. 10. Specifically, Appellants contend the alignment processing section in
Kawashima derives alignment information from the second individual's
vocal characteristics, but does not align the voice of the second individual to
match the singing voice of the first individual. Br. 10 ( citing Kawashima,
col. 18, 11. 24--31 ). Rather, Appellants assert, Kawashima is generally
directed to "a voice converter for assimilating a user voice to be processed to
9
Appeal2018-000374
Application 14/383,341
a different target voice." Kawashima, col. 1, 11. 9-11. Figure 1 of
Kawashima, as relied on by the Examiner (see, e.g., Ans. 7-8) is illustrative
and is reproduced below:
iO : VOICE CONVERTER 14
15
16
(KEY ANO TEMPO
INFORMATION}
FIG.. 1
12
Figure 1 of Kawashima is a high-level block diagram of Kawashima's
disclosed voice converter (10). Kawashima, col. 6, 11. 59-61. Kawashima
discloses the voice converter as applied to a karaoke apparatus is "capable of
performing imitation of a target singer." Kawashima, col. 7, 11. 49-52. In
the disclosed embodiment, Kawashima describes that data, such as pitch and
amplitude temporal changes, are extracted from the singing voice of a target
singer and such information is stored in Target Behavior Data storage (15).
Kawashima, col. 9, 11. 22-38. Additionally, "spectrum information
corresponding to respective phonemes of the target singer to be imitated" are
10
Appeal2018-000374
Application 14/383,341
stored in Target Phonemic Dictionary (19). Kawashima, col. 9, 11. 39--46.
Kawashima further describe the singing signal input section ( 11) inputs a
voice signal generated by a karaoke signer (i.e., the voice to be modified by
the target characteristic data). Kawashima, col. 17, 11. 55-58. The input
voice signal is analyzed in the Recognition Feature Analysis Section (12)
that calculates characteristic vectors of the signal. Kawashima, col. 1 7, 11.
59-67. Additionally, the SMS Analysis Section (13) also analyzes the input
singing signal to generate, inter alia, frame data. Kawashima, col. 18,
11. 10-14. Kawashima discloses the Alignment Processing Section (18):
finds respective parts of the music piece sung by the karaoke
singer using Viterbi algorithm on the basis of various
characteristic vectors VC outputted from the recognition feature
analysis section 12, HMM [(Hidden Markov Model)] of
respective phonemes from the recognition phonemic dictionary
14, and the phonemic notation information with duration
included in the target behavior data.
Kawashima, col. 18, 11. 24--31. On the basis of the alignment information
(i.e., the output from the Alignment Processing Section (18)), Target
Decoder Section (20) calculates frame information ( a pitch, an amplitude,
and a spectrum shape) of the target singer. Kawashima, col. 18, 11. 42--49.
Morphing Processing Section (21) morphs the singing signal (based on
target frame data, frame data, and likeness data; Conversion Processing
Section (22) transforms morphing frame data into conversion frame data;
and SMS Synthesizing Section (23) outputs a synthesized output signal.
Kawashima, col. 18, 1. 50-col. 19, 1. 5.
Contrary to Appellants' arguments, we agree with the Examiner that
Kawashima discloses aligning the voice of the second individual (i.e., the
singing signal input) to match the singing voice of the first individual (i.e.,
11
Appeal2018-000374
Application 14/383,341
the target singer). See Ans. 8. As described, the characteristics of a singing
voice of a first individual are derived and stored at least in Target Behavior
Data (15) and Target Phonemic Dictionary (19). The voice of a second
individual (i.e., the singing signal input) is analyzed (12) to obtained vocal
characteristics, which are aligned with the singing voice data of the first
individual (i.e., target singer) as part of the vocal characteristic modification
process of the singing signal (i.e., the voice of the second individual).
For the reasons discussed supra, we are unpersuaded of Examiner
error. Accordingly, we sustain the Examiner's rejection of independent
claim 1. Additionally, we sustain the Examiner's rejection of claims 3---6
and 8, which depend directly or indirectly therefrom and were not argued
separately. See Br. 11; see also 37 C.F.R. § 4I.37(c)(l)(iv).
b. Haupt ( claim 19)
Appellants contend Haupt fails to disclose "synthesizing a singing
voice for the second individual in response to the converted spectral
parameter sequence and the pitch contour information of the singing voice of
the first individual," as recited in independent claim 19. Br. 11-12. Rather
than synthesizing a singing voice for the second individual, Appellants assert
Haupt discloses adapting a pitch of the first individual. Br. 11-12 ( citing
Haupt ,r,r 42--43). Additionally, Appellants argue such pitch adaptation is
not in response to the converted spectral parameter sequence and pitch
contour information. Br. 12.
Haupt is generally directed to converting a spoken voice to a singing
voice sung in the manner of a target singer. Haupt, Title. In a disclosed
embodiment, a user speaks the lyrics of a song and the captured voice is
12
Appeal2018-000374
Application 14/383,341
sampled and processed to "cause[] the user's voice to sound as if the user
were singing the song with the same pitch and timing as the original artist."
Haupt ,r 9. Haupt describes major blocks of the voice synthesis process:
(i) voice analysis of the spoken fragments/words of the source speaker;
(ii) voice analysis of the target singer; (iii) re-estimation of parameters of the
source speaker to match the target singer; and (iv) re-synthesis of the source
model to make a singing voice. Haupt ,r 30. Haupt describes the voice
analysis system is a "low-order parameterization of speech" that uses a
linear prediction coefficient (LPC) analysis to break the voice into two
separate components-a pitch signal and a filter signal. Haupt ,r 32. Haupt
further describes obtaining pitch estimates for the speaker's and singer's
voice data using, for example, autocorrelation methods. Haupt ,r 33. Haupt
also discloses using timing information from (i.e., the length of a triphone)
to scaled the spoken utterance to be the same length as the sung triphone.
Haupt ,r 35. Haupt further discloses the pitch of the singer "is transposed to
the speaker's voice", using a pitch multiplication constant. Haupt ,r 3 6.
"Deterministic and stochastic sections of speech are transposed from the
singer to the speaker." Haupt ,r 36. Haupt also describe that a singing voice
model may be added to change a characteristic of speech into singing.
Haupt ,r 37. In addition to matching transitions between boundaries between
triphones, Haupt describes an energy-shaping component to match the
amplitude shape. Haupt ,r,r 39--40. Using different variable and filter
coefficients, Haupt describes a singing voice sample may be synthesized.
Haupt ,r,r 41--42. Additionally, using the timing information of the
triphones, "the resultant speech sample after voice synthesis 86 is then
placed in a signal timed in such a manner that the sung voice and the newly
13
Appeal2018-000374
Application 14/383,341
formed sample occur at the exact same point in the song." Haupt. ,r 42.
Haupt describes that the resulting track "will be singing in a speaker's voice
in the manner of a target singer." Haupt ,r 42.
Contrary to Appellants' arguments, we agree with the Examiner that
Haupt discloses synthesizing a singing voice for a second individual (i.e., the
speaker) in response to the converted spectral parameter sequence and the
pitch contour information of the singing voice of the first individual.
Ans. 9-10; see also Haupt ,r,r 30-43. Specifically, the Examiner finds, and
we agree, Haupt discloses that filter coefficients for the singer and source are
transformed to the frequency domain and a scaling factor is calculated to
match the amplitude shape of the source speaker to the target. Ans. 9 ( citing
Haupt ,r,r 40-42). The filter coefficients and scaling factor correspond to the
claims converted spectral parameters used to transform the source signal to a
target singing voice. Ans. 9. Further, as the Examiner explains, Haupt
discloses transposing the pitch of the singing voice to a speaker voice.
Ans. 9-10 (citing Haupt ,r,r 36, 43). The Examiner correctly identifies that
Haupt describes, as part of the synthesis of a singing voice for a second
individual, the pitch of the singing voice is transposed to the second
individual. See Haupt ,r 36. Thus, we agree with the Examiner that Haupt
discloses "synthesizing a singing voice for the second individual in response
to the converted spectral parameter sequence and the pitch contour
information of the singing voice of the first individual," as recited in
independent claim 19.
For the reasons discussed supra, we are unpersuaded of Examiner
error. Accordingly, we sustain the Examiner's rejection of independent
claim 19.
14
Appeal2018-000374
Application 14/383,341
Rejections under 35 U.S.C. § 103
a. Claim 2
Claim 2 depends from claim 1 and recites, inter alia, "the voice of the
second individual is speech." Appellants argue Kawashima discloses
aligning vocal characteristics of the first individual's singing voice to the
second individual's (i.e., the karaoke singer's) singing voice-i.e., not to the
spoken voice (speech) of the second individual. Br. 12-13. Further,
Appellants argue Haupt discloses aligning a spoken utterance of a second
individual with a first individual's singing utterance. Br. 13. Appellants
assert Haupt discloses performing the alignment using timing information
and the pitch-synchronous overlap and add (PSOLA) method, whereas
Appellants' Specification discloses characteristics may include pitch contour
information, alignment information and spectral parameter sequence
information. Br. 13.
We are not persuaded of Examiner error. As the Examiner explains,
and as discussed above, Haupt describes aligning the spoken voice (i.e.,
utterances or speech) with the singing voice of a target (i.e., first individual).
Ans. 10-11 ( citing Haupt ,r,r 31--40, 48). Appellants' discussion that the
Specification discloses certain characteristics as opposed to Haupt' s use of
timing information and PSOLA to perform alignment is also unavailing at
least because Appellants' characteristics of pitch contour information,
alignment information and spectral parameter sequence information are not
recited in either claim 1 or claim 2. See In re Van Geuns, 988 F.2d 1181,
1184 (Fed. Cir. 1993) (limitations are not to be read into the claims from the
Specification). Additionally, as the Examiner explains (Ans. 10) and as
15
Appeal2018-000374
Application 14/383,341
discussed above, Kawashima discloses aligning a singing voice of a second
individual to match the singing voice of a first individual. In the rejection,
the Examiner explains it would have been obvious to an ordinarily-skilled
artisan to modify the singing voice of the second individual ( as described by
Kawashima) with the spoken voice of a second individual (as described by
Haupt) to synthesize a singing voice using less effort. Final Act. 13-14;
Ans. 11-12. We agree. Accordingly, we sustain the Examiner's rejection
under 35 U.S.C. § 103(a) of claim 2.
b. Claims 9, 10, and 14--17
Appellants also argue the Examiner erred in finding Cen teaches or
reasonably suggests "deriving a template of first speech characteristics and
first singing characteristics in response to a first individual's speaking voice
and singing voice," as recited in independent claim 9. Br. 14--15. Rather,
Appellants assert Cen teaches using "a pair of template speaking and singing
voices," not speech and singing characteristics. Br. 15.
In response, the Examiner finds Cen teaches template-based Speech-
to-Singing (STS) synthesis "trains the transformation models by analyzing a
pair of template speaking and singing voices that are usually [ derived] from
a professional singer." Cen, Introduction; Ans. 14 (emphasis omitted). In
other words, "the template-based models are first obtained from a first
individual." Ans. 14. Further, Cen describes that in the learning stage of the
STS synthesis system template singing and speaking voices are further
analyzed to derive transformation models based on acoustic features such as
pitch contour, phoneme duration, and spectrum. Cen, Section II; Ans. 14.
As the Examiner explains, "the acoustic features (characteristics) are
16
Appeal2018-000374
Application 14/383,341
extracted [(from the speaking and singing voices)] to form a template[]-
based transform function ( obtained from a first individual), and a new
utterance ( obtained from a second individual) is converted into a singing
voice by modifying the acoustic features using the transformation model."
Ans. 14 ( emphasis added). In other words, the Examiner finds Cen starts
with a pair of template singing and speaking voices that are then analyzed to
extract acoustic features (i.e., speech and singing characteristics). Thus, the
speech and singing characteristics are template based and derived from a
first individual's speaking voice and singing voice. Appellants do not rebut
persuasively the Examiner's findings.
Additionally, Appellants contend the Examiner erred in finding the
combination of Haupt and Cen teaches or reasonably suggests "aligning
acoustic features of the second individual's approximated singing voice to
the template of the first speech characteristics, the first singing
characteristics and the second speech characteristics to generate the second
individual's synthesized singing voice." Br. 15. In particular, Appellants
argue the utterances disclosed in Haupt are not the acoustic features and
speech/singing characteristics recited in claim 9. Br. 15. Further,
Appellants again assert Cen does not disclose speech and singing
characteristics. Br. 15.
The Examiner provides a comprehensive response detailing how each
of Haupt and Cen teaches or reasonably suggests the alignment of the
claimed elements. See Ans. 15-19. Regarding Haupt, the Examiner finds,
and we agree (see discussion of Haupt supra), Haupt describes extracting
acoustic features and characteristics from a speech signal (i.e., utterance) to
be converted to a synthesized singing signal. Ans. 19; see also Haupt ,r 33
17
Appeal2018-000374
Application 14/383,341
( e.g., extracting pitch estimates). Additional characteristics extracted or
determined include alignment information (e.g., triphone boundaries) and
spectral parameter sequence information ( e.g., filter coefficients). See Haupt
,r,r 38--40; Ans. 19. Regarding Appellants' argument the teaching of Cen,
for similar reasons discussed earlier in this section, we are unpersuaded of
Examiner error.
For the reasons discussed supra, we are unpersuaded of Examiner
error. Accordingly, we sustain the Examiner's rejection of independent
claim 9. For similar reasons, we also sustain the Examiner's rejection of
claims 10 and 14--17, which depend directly or indirectly therefrom and
were not argued separately. See Br. 16; see also 37 C.F.R. § 4I.37(c)(l)(iv).
c. Claims 7, 11-13, and 18
Regarding dependent claims 7, 11-13, and 18, Appellants assert the
additionally cited references fail to overcome the alleged deficiencies related
to the independent claims. Br. 13-14, 16. Because we do not find the
Examiner's rejection of the independent claims to be in error, we similarly
do not find the rejection of these claims to be in error. Accordingly, we
sustain the Examiner's rejections of claims 7, 11-13, and 18.
DECISION
We affirm the Examiner's decision rejecting claims 1-19 under
35 U.S.C. § 101.
We affirm the Examiner's decision rejecting claims 1, 3-6, 8, and 19
under 35 U.S.C. § 102(a).
18
Appeal2018-000374
Application 14/383,341
We affirm the Examiner's decision rejecting claims 2, 7, and 9-18
under 35 U.S.C. § 103(a).
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(l)(iv). See
37 C.F.R. § 41.50(±).
AFFIRMED
19