Ex Parte Darroudi et al

Board of Patent Appeals and Interferences

Jun 26, 2012

10336668 (B.P.A.I. Jun. 26, 2012)

UNITED STATES PATENT AND TRADEMARK OFFICE
____________

BEFORE THE BOARD OF PATENT APPEALS
AND INTERFERENCES
____________

Ex parte KHOSRO DARROUDI and BRIAN R. MEARS
____________

Appeal 2010-003328
Application 10/336,668
Technology Center 2600
____________

Before JOHN A. JEFFERY, THOMAS S. HAHN, and JOHN A. EVANS,
Administrative Patent Judges.

EVANS, Administrative Patent Judge.

DECISION ON APPEAL
Appellants appeal under 35 U.S.C. § 134(a) from the Examiner’s rejection
of claims 1-30. We have jurisdiction under 35 U.S.C. § 6(b). We REVERSE.

Appeal 2010-003328
Application 10/336,668

2

STATEMENT OF THE CASE
Appellants describe their invention as follows:

A system for encoding speech includes a speech encoder (106, Figure l), a
speech recognizer (110), and a difference encoder (l08). When the speech
recognizer (110) recognizes a word, phoneme or feature within an input
speech signal (122), the difference encoder (108) calculates the differences
between speech parameters (140, 142) derived by the speech encoder (106)
and speech parameters (146, 148) derived by the speech recognizer (110).
The difference encoder (l08) quantizes the differences (128), which replace
corresponding encoder-derived parameters to be transmitted over a channel
(130). In one embodiment, the difference encoder representation (128) of the
speech parameters consumes fewer bits than the encoder-derived
representation (124). Accordingly, the resulting bandwidth consumed by a
single channel can be decreased.

(Spec. 37, Abstract.)

Claim 1 is illustrative:
1. A method for encoding speech comprising:

processing an input speech signal using an encoder, resulting in
a compressed encoder representation of the input speech signal;

if a speech recognizer identifies a corresponding dictionary
speech element, which approximates the input speech signal,

determining a compressed recognizer representation of the
corresponding dictionary speech element,

calculating one or more differences between the compressed
encoder representation and the compressed recognizer representation,

compiling compressed speech information that includes
representations of the one or more differences; and

Appeal 2010-003328
Application 10/336,668

3

the method further comprising, if the speech recognizer does
not identify a corresponding dictionary speech element, compiling the
compressed speech information to include the compressed encoder
representation of the input speech signal, and not to include the one or
more differences.

THE REJECTIONS
I. The Examiner rejected claims 1, 10-13, 19, 24, 25, and 28 under 35 U.S.C.
§ 103(a) as obvious over Li (US 5,305,421, issued Apr. 19, 1994) and
Ittycheriah (US 6,119,086, issued Sep. 12, 2000, filed Apr. 28, 1998).
Ans. 3.
II. The Examiner rejected claims 26 and 27 under 35 U.S.C. § 103(a) as
obvious over Li and Ittycheriah, in view of Appellants’ admitted prior art.
Ans. 6.
III. The Examiner rejected claims 2-6, 14-17, 20-23, 29, and 30 under
35 U.S.C. § 103(a) as obvious over Li, Ittycheriah, and Ozawa (US
5,487,128, issued. Jan. 23, 1996). Ans. 7.
IV. The Examiner rejected claims 7, 8 and 18 under 35 U.S.C. § 103(a) as
obvious over Li, Ittycheriah, Ozawa, and Maung (US 5,924,062, issued
July 13, 1999). Ans. 13.
V. The Examiner rejected claim 9 under 35 U.S.C. § 103(a) as obvious over
Li, Ittycheriah, and Cohen (US 6,219,643 B1, issued Apr. 17, 2001). Ans.
15.

Appeal 2010-003328
Application 10/336,668

4

THE REJECTION OF CLAIMS 1, 10-13, 19, 24, 25, AND 28
ISSUE
Appellants argue: “In the present case claims 1, 10-13, 19, 24-25 and 28 are
not obvious under 35 U.S.C. § 103(a) because the combined teaching of Li and
Ittycheriah does not teach or suggest every element of any of the rejected claims.”
(App. Br. 21-22). The issue is whether the Examiner failed to locate each claimed
element in the prior art.1

ANALYSIS
The Examiner’s prima facie case initially analyzes Claim 1 into a first set of
limitations relating to the case where a speech recognizer routine determines that a
spoken word is known and a second set of limitations relating to the case where a
speech recognizer routine determines that a spoken word is not known. The
Examiner then separately applies Li and Ittycheriah to the two cases:
The combination of Li and Ittycheriah presented in the rejection is
that a speech recognizer first determines if the spoken word is a
known word. If it is known then it performs processing in the way Li
describes.
(Ans. 21).
In the event that the speech recognizer determines that the spoken word
is unknown (as described by Ittycheriah), then the speech signal is
forwarded to Ittycheriah's processor for generating an output that is then
sent to the decoder.
(Ans. 21-22).
Therefore, the processes themselves used to process the input speech
are independent of each other, and yet they are both used (i.e.,
combined) to form a system that is capable of handling any speech

1 Appellants present further allegations of error. See App. Br. 18-21. However,
because we find the present issue to be dispositive of all the issues before us, we do
not reach the merits of those other contentions.
Appeal 2010-003328
Application 10/336,668

5

including speech that includes both known words and unknown
words.
(Ans. 22).
Appellants argue that where a spoken word is recognized, the separate
application of Li results in an incomplete mapping of the claim limitations:
If the speech recognizer identifies a dictionary speech element that
approximates the input speech signal, the claims call for calculating
differences between the compressed encoder representation of the
input speech signal and a compressed recognizer representation of the
input speech signal. The claims then call for compiling speech
information that includes representations of the differences between
the compressed encoder representation and the compressed recognizer
representation.
(App. Br. 22).
Li teaches sending recognized word data over first
communication path and side information (comprising the median
pitch, the average pitch slope, amplitude and duration of words) over
a second communication path. Parameters of recognized words are
compared with dictionary parameters or word template parameters of
the recognized words stored in the continuous speech recognition
system. The differences between the two are quantized and coded into
one of 128 (7 bits) different codes which are transmitted over the
second communication path. No such coding is performed on the
recognized word data transmitted over the first communication path.
(App. Br. 22-23).
Thus, according to Li, the only place where compressed speech information
is compiled that includes representations of the one or more differences between
recognized word parameters and stored dictionary parameters is within the side
information encoding signal path.
(App. Br. 23).
The Examiner replies to Appellants’ additional arguments relating to the
combination of Li with Ittycheriah for the case where an input word is not
recognized. However, the Examiner does not reply to Appellants’ contentions,
supra, that where an input word is recognized, the method of Li, as separately
Appeal 2010-003328
Application 10/336,668

6

applied by the Examiner, fails to teach compiling representations of the differences
between the compressed encoder representation and the compressed recognizer
representation of recognized words. (See Br. 23). Moreover, our independent
review of the Examiner’s rejection fails to reveal a compilation comprising
representations of the differences between a compressed encoder representation
and a compressed recognizer representation of recognized words. Therefore, even
if the references were combinable as the Examiner proposes, the references still
fail to teach or suggest all limitations of independent claim 1.
The Examiner acknowledges that Appellants argue that “the references fail
to teach every element of any of the claims.” (Ans. 33). However, the Examiner
focuses on Appellants’ argument the Li and Ittycheriah are not properly
combinable, but fails to address Appellants’ contention that Li does not teach every
element where a word is recognized. (See Ans. 33-36).
In view of the foregoing, we are persuaded that the Examiner erred in
rejecting (1) independent claim 1; (2) independent claims 13, 19, 25, and 28 which
recite commensurate limitations; and (3) dependent claims 10-12 and 24 for similar
reasons.
Claims 2-9, 14-18, 20-23, 26, 27, 29, and 30 were rejected as obvious over
Li and Ittychariah in view of additional prior art. Because the Examiner did not
cite the additional art as teaching compiling representations of the “differences
between the compressed encoder representation and the compressed recognizer
representation” (see claim 1) of recognized words, we are persuaded that the
Examiner erred in rejecting claims 2-9, 14-18, 20-23, 26, 27, 29, and 30.

Appeal 2010-003328
Application 10/336,668

7

CONCLUSIONS
The Examiner erred in rejecting (1) claims 1, 10-13, 19, 24, 25, and 28 under
35 U.S.C. § 103(a) as obvious over Li and Ittycheriah and (2) claims 2-9, 14-18,
20-23, 26, 27, 29, and 30 as obvious over Li and Ittycheriah in view of secondary
references.

DECISION
The Examiner’s decision rejecting claims 1-30 is reversed.

REVERSED