Ex Parte Veliu et al

Patent Trial and Appeal Board

Dec 16, 2014

11301949 (P.T.A.B. Dec. 16, 2014)

UNITED STATES PATENT AND TRADEMARKOFFICE
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
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR ATTORNEY DOCKET NO. CONFIRMATION NO.
11/301,949 12/13/2005 Shpetim S. Veliu GP-306822 (2760/260) 9862
60770 7590 12/17/2014
General Motors Corporation
c/o REISING ETHINGTON P.C.
P.O. BOX 4390
TROY, MI 48099-4390
EXAMINER
ORTIZ SANCHEZ, MICHAEL
ART UNIT PAPER NUMBER
2658
MAIL DATE DELIVERY MODE
12/17/2014 PAPER
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.
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
____________________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
____________________

Ex parte SHPETIM S. VELIU, HITAN S. KAMDAR, ANTHONY J.
SUMCAD, RUSSELL A. PATENAUDE, BRAD T. REESER,
RATHINAVELU CHENGALVARAYAN, SCOTT M. PENNOCK, and
TIMOTHY J. GROST
____________________

Appeal 2012-004192
Application 11/301,949
Technology Center 2600
____________________

Before ST. JOHN COURTENAY III, THU A. DANG,
and LARRY J. HUME, Administrative Patent Judges.

DANG, Administrative Patent Judge.

DECISION ON APPEAL

Appeal 2012-004192
Application 11/301,949

2
I. STATEMENT OF THE CASE
Appellants appeal under 35 U.S.C. § 134(a) from the Final Rejection
of claims 1, 2, 4, 6–9, 11, 12, 21–24, and 26–34. Claims 3, 5, 10, 13–20,
and 25 have been canceled. The Examiner indicates that claim 31 would be
allowable if rewritten in independent form. (Ans. 4). We have jurisdiction
under 35 U.S.C. § 6(b).
We affirm.

A. INVENTION
According to Appellants, the invention generally relates to
customizing speech recognition according to speech regions based on
instances of failed speech recognition within a mobile vehicle
communication system (Spec. ¶ 1).

B. ILLUSTRATIVE CLAIM
Claim 1 is exemplary:
1. A method of customizing speech recognition in a mobile
vehicle communication system, comprising the steps of:
receiving a speech input from a user in a vehicle;
comparing the speech input with a standard voice recognition
set using a voice recognition algorithm;
determining that a speech recognition failure has occurred if the
speech input is not recognized;
in response to the determined speech recognition failure,
marking the speech input with an identifier that associates the speech
input with the user and that designates a user record including speech
misrecognitions;
sending the marked speech input to a call center;
using the identifier to access, in a database at the call center, a
user-specific, voice recognition set alternative to the standard voice
Appeal 2012-004192
Application 11/301,949

3
recognition set and uniquely created for the user based on previously
determined speech patterns; and
sending the alternative, user-specific, voice recognition set to
the telematics unit.

C. REJECTIONS
The prior art relied upon by the Examiner as evidence in rejecting the
claims on appeal is:
Bareis US 2003/0087675 A1 May 8, 2003
Everhart US 6,587,824 B1 Jul. 1, 2003
Lenane US 2004/0107097 A1 Jun. 3, 2004
Soufflet US 2005/0102142 A1 May 12, 2005
Runge US 2006/0223512 A1 Oct. 5, 2006
Chutorash US 2007/0005368 A1 Jan. 4, 2007

Claims 1, 2, 6, 8, 9, 21, 24, 27–29, and 32–34 stand rejected under
35 U.S.C. § 103(a) as being unpatentable over Runge, Everhart, and
Soufflet.
Claims 30 and 31 stand rejected under 35 U.S.C. § 103(a) as being
unpatentable over Runge, Everhart, Chutorash, and Soufflet.
Claims 4, 5, 7, 11, 12, 23, and 25 stand rejected under 35 U.S.C.
§ 103(a) as being unpatentable over Runge, Everhart, Soufflet, and Lenane.
Claim 22 stands rejected under 35 U.S.C. § 103(a) as being
unpatentable over Runge, Everhart, Soufflet, and Bareis.

II. ISSUE
The dispositive issue before us is whether the Examiner has erred in
finding the combination of Runge, Everhart, and Soufflet teaches or suggests
the following contested limitations: “comparing the speech input with a
standard voice recognition set,” “ determining that a speech recognition
Appeal 2012-004192
Application 11/301,949

4
failure has occurred if the speech input is not recognized,” and “in response
to the determined speech recognition failure, marking the speech input with
an identifier that associates the speech input with the user and that
designates a user record including speech misrecognitions” wherein the
identifier is used to access “a user-specific voice recognition set alternative .
. . uniquely created for the user” (claim 1, emphasis added).

III. FINDINGS OF FACT
The following Findings of Fact (FF) are shown by a preponderance of
the evidence.
Soufflet
Soufflet discloses detecting non-recognized expressions and
generating data for correcting the language model based on the non-
recognized expressions for subsequent recognition (Abst.), wherein Soufflet
takes into account language styles or turns of phrases specific to certain
users (¶ 35). The data representative of the unrecognized expressions
comprise a compressed voice recording ([0038]), and the voice data is sent
to the source for fine analysis at the server level ([0039]), wherein at least
one item of information transmitted to the server ([0040]) includes
information relating to the speaker who has uttered an unrecognized
expression (¶ 42).

IV. ANALYSIS
Appellants contend “Runge does not teach or suggest comparing
speech input with a standard voice recognition set or accessing and
downloading an alternative voice recognition set” (App. Br. 8–9, emphasis
Appeal 2012-004192
Application 11/301,949

5
omitted). Appellants also contend “Everhart does not teach or suggest
accessing a voice recognition set in a call center database, and certainly not a
voice recognition set alternative to a standard voice recognition set” or
“locating an alternative voice recognition set in response to a failure to
recognize speech input with a first voice recognition set” (App. Br. 10,
emphasis omitted). Further, Appellants contend Soufflet’s “correcting a
phonetic set and/or a language model is not the same as locating a voice
recognition set alternative to a standard voice recognition set” and “updating
of language models is not the same as Applicant’s claimed user-specific
voice recognition set uniquely created for a user” (App. Br. 11). Appellants
then contend “there is no valid reason” to combine the references, and even
if the references could be combined, the combined teachings would not
include Appellants’ invention “absent hindsight” (App. Br. 14).
Although Appellants contend none of the references teaches or
suggests the claimed invention (App. Br. 8–11), the test for obviousness is
not what the references show individually but what the combined teachings
would have suggested to one of ordinary skill in the art. See In re Merck &
Co., Inc., 800 F.2d 1091, 1097 (Fed. Cir. 1986).
The Examiner points out Soufflet is relied upon for teaching
“receiving data representing the unrecognized expression” and “sending to
the remote terminal correcting information obtained on the basis of an
analysis of the data received at the server, the correcting information
allowing the correcting of the language mode allowing future recognition of
the unrecognized expressions” (Ans. 23). The Examiner finds, because
“Soufflet teaches implementing a voice recognition at a terminal,” Soufflet
discloses “a standard voice recognizer is implemented” (Ans. 24).
Appeal 2012-004192
Application 11/301,949

6
According to the Examiner, “[t]he broadest reasonable interpretation of an
alternative to a standard voice recognition set is any recognition set which
has at least one different item” and thus Soufflet discloses “receiving an
alternative recognition set” (Ans. 26).
The Examiner further finds “Soufflet teaches that data representative
of the unrecognized expressions comprise a voice recording representative
of parameters descriptive of the acoustic signal” and “information relating to
the speaker” are sent with the information which “are identifiers associated
with the misrecognized expression” (Ans. 27). That is, “Soufflet teaches
identifier data including information of content of use of the voice
recognition process when an expression is not recognized” wherein
“information relating to the user who uttered the unrecognized expression is
sent” therefore “a user record is created” (Ans. 29).
The Examiner then finds it would have been obvious to “combine the
Runge method of updating a speech recognition algorithm in a mobile
device . . . with the Soufflet method of speech recognition which updates
language models” for “the benefit enabling subsequent recognition of at
least some of the non-recognized expressions” (Ans. 31–32).
We find no error with the Examiner’s underlying factual findings and
ultimate legal conclusion of obviousness of the claims over the combined
teachings and suggestions of the cited references.
To determine whether the claims would have been obvious over the
combined teachings, we give the claims their broadest reasonable
interpretation consistent with the Specification. See In re Morris, 127 F.3d
1048, 1054 (Fed. Cir. 1997). We note claim 1 does not define “standard
voice recognition set” and “alternative, user-specific, voice recognition set”
Appeal 2012-004192
Application 11/301,949

7
other than the user-specific voice recognition set is alternative to the
standard voice recognition set. Further, the Specification does not provide
any specific definition for the terms.
Soufflet discloses detecting non-recognized expressions and
generating data for correcting the language model based on the non-
recognized expressions (FF). That is, Soufflet discloses detecting using an
uncorrected language model, wherein upon detection of non-recognized
expressions in comparison the expressions in the uncorrected language
model, a corrected language model is generated using correction data (id.).
We agree with the Examiner’s broad but reasonable interpretation that
Soufflet’s voice recognition “at a terminal” comprises “a standard voice
recognizer” (Ans. 24), and that “an alternative to a standard voice
recognition set is any recognition set which has at least one different item”
wherein Soufflet’s corrected language model comprises such “alternative
recognition set” (Ans. 26). Thus, we agree with the Examiner, Soufflet at
least suggests the contested limitations: “comparing the speech input with a
standard voice recognition set,” “determining that a speech recognition
failure has occurred if the speech input is not recognized,” and creating an
alternative voice recognition set “in response to the determined speech
recognition failure,” as required by claim 1.
Further, Soufflet takes into account language styles or turns of phrases
specific to certain users, wherein data representative of the unrecognized
expressions is sent to the server comprising a compressed voice recording
and information relating to the speaker who has uttered the unrecognized
expression (FF). That is, Soufflet discloses correcting the language model
for subsequent recognition, wherein in response to determining an
Appeal 2012-004192
Application 11/301,949

8
expression is unrecognized (speech recognition failure), marking the voice
recording with information that associates the voice recording with the
particular user wherein the corrected language model is uniquely created for
the user to take into account for the different language styles or turns of
phrases previously determined (id.). Contrary to Appellants’ contention the
references “would not include, absent hindsight through Appellants’
disclosure, locating an alternative voice recognition set . . . ,” (App. Br. 14),
Soufflet at least suggest a corrected language model that is specific to a
particular user and uniquely created for the user based on previously
predetermined speech patterns (FF).
We find no error with the Examiner’s finding that Soufflet teaches or
at least suggests “information relating to the speaker” is sent with the
information which “are identifiers associated with the misrecognized
expression” (Ans. 27), wherein “identifier data” includes “information of
content of use of the voice recognition process when an expression is not
recognized” and wherein “a user record is created” therewith (Ans. 29).
That is, Soufflet at least suggests an “identifier” that marks the voice
recording and designates the speaker’s informational record which includes
the unrecognized expressions which is then used to access the corrected
language model that is created for the user, taking into account language
styles or turns of phrases specific to the user (FF).
Thus, we agree with the Examiner Soufflet at least suggests “in
response to the determined speech recognition failure, marking the speech
input with an identifier that associates the speech input with the user and that
designates a user record including speech misrecognitions” wherein the
Appeal 2012-004192
Application 11/301,949

9
identifier is used to access “a user-specific, voice recognition set alternative
. . . uniquely created for the user” as required by claim 1.
Furthermore, even assuming, arguendo, Appellants are correct that
the combined teachings do not “include” an “identifier” that “designates a
user record including speech misrecognitions,” or “locating an alternative
voice recognition set that is specific to a particular user and uniquely created
for the user” (App. Br. 14, emphasis omitted), Appellants have provided no
evidence that modifying Soufflet’s system that takes into account a user’s
unique speech pattern (FF) to include a user record which includes speech
misrecognition or an alternative voice recognition set that is specific to the
user was “uniquely challenging or difficult for one of ordinary skill in the
art,” Leapfrog Enterprises, Inc. v. Fisher-Price, Inc., 485 F.3d 1157, 1162
(Fed. Cir. 2007). Neither have Appellants presented evidence that this
incorporation yielded more than expected results.
Rather, we find that Appellants’ invention is simply a combination of
known teachings that would have realized a predictable result. The Supreme
Court has determined the conclusion of obviousness can be based on the
interrelated teachings of multiple patents, the effects of demands known to
the design community or present in the marketplace, and the background
knowledge possessed by a person having ordinary skill in the art. KSR Int’l
Co. v. Teleflex Inc., 550 U.S. 398, 418 (2007). The skilled artisan is “a
person of ordinary creativity, not an automaton.” Id. at 420–21.
We also find no error with the Examiner’s proffered rationale to
“combine the Runge method of updating a speech recognition algorithm in a
mobile device . . . with the Soufflet method of speech recognition which
updates language models” for “the benefit of enabling subsequent
Appeal 2012-004192
Application 11/301,949

10
recognition of at least some of the non-recognized expressions” (Ans. 31–
32).
Although Appellants also contend “there is no valid reason” to
combine the references (App. Br. 14), we find the Examiner provided
sufficient articulated reasoning with some rational underpinning to support
the legal conclusion of obviousness. (Ans. 31–32). Further, as to
Appellants’ contention the combined teachings would not include
Appellants’ invention “absent hindsight” (App. Br. 14), although we are
fully aware that hindsight bias often plagues determinations of obviousness,
Graham v. John Deere Co., 383 U.S. 1, 36 (1966), we are also mindful the
Supreme Court has clearly stated the “combination of familiar elements
according to known methods is likely to be obvious when it does no more
than yield predictable results.” KSR, 550 U.S. at 416. “In determining
whether the subject matter of a patent claim is obvious, neither the particular
motivation nor the avowed purpose of the patentee controls.” Id. at 419.
Accordingly, we find no error in the Examiner’s rejection of
independent claim 1 over Runge, Everhart, and Soufflet. Appellants do not
provide substantive arguments for independent claims 8 and 21, and claims
2, 6, 9, 24, 27–29, and 32–34 depending respectively from independent
claims 1, 8, and 21 (App. Br. 14). Accordingly, claims 2, 6, 8, 9, 21, 24, 27–
29, and 32–34 fall with claim 1.
As for claims 30 and 31, although Appellants contend “Chutorash
does not teach or suggest that an input nametag is compared to anything and
certainly not misrecognized speech input” (App. Br. 16), Appellants are
reminded the test for obviousness is not what Chutorash shows individually
but what the combined teachings would have suggested to one of ordinary
Appeal 2012-004192
Application 11/301,949

11
skill in the art. Nevertheless, we find no error with the Examiner’s finding
“Chutorash teaches that a speech recognition function recognizes a word”
wherein “the speech recognition performed identifies and ranks a plurality of
recognitions for each word in the spoken command” (Ans. 32).
Accordingly, we also find no error in the Examiner’s rejection of
claims 30 and 31 over Runge, Everhart, and Soufflet in further view of
Chutorash. Appellants do not provide substantive arguments for claims 4, 5,
7, 11, 12, 22, 23, and 25 (App. Br. 17). Accordingly, we also affirm the
rejection of claims 4, 5, 7, 11, 12, 23, and 25 over Runge, Everhart, and
Soufflet in further view of Lenane, and the rejection of claim 22 over Runge,
Everhart, and Soufflet in further view of Bareis.

V. CONCLUSION AND DECISION
The Examiner’s rejections of claims 1, 2, 4, 6–9, 11, 12, 21–24, and
26–34 under 35 U.S.C. § 103(a) are affirmed.
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(1)(iv).

AFFIRMED

gvw