Ex Parte Nilsson et al

Patent Trial and Appeal Board

Aug 31, 2017

12635235 (P.T.A.B. Aug. 31, 2017)

United States Patent and Trademark Office
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.
12/635,235 12/10/2009 Mattias Nilsson 335483-US-CIP 8963
69316 7590 09/05/2017
MICROSOFT CORPORATION
ONE MICROSOFT WAY
REDMOND, WA 98052
EXAMINER
ORTIZ SANCHEZ, MICHAEL
ART UNIT PAPER NUMBER
2658
NOTIFICATION DATE DELIVERY MODE
09/05/2017 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):
u sdocket @ micro soft .com
chriochs @microsoft.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte MATTIAS NILSSON, SOREN VANG ANDERSON,
and KOEN BERNARD VOS
Appeal 2017-005466
Application 12/635,2351
Technology Center 2600
Before CARLA M. KRIVAK, HUNG H. BUI, and JON M. JURGOVAN,
Administrative Patent Judges.
BUI, Administrative Patent Judge.
DECISION ON APPEAL
Appellants seek our review under 35 U.S.C. § 134(a) from the
Examiner’s Final Rejection of claims 1—5 and 7—24, which are all the claims
pending in the application. We have jurisdiction under 35 U.S.C. § 6(b).
We AFFIRM-IN-PART.2
1 According to Appellants, the real party in interest is Skype. Br. 3.
2 Our Decision refers to Appellants’ Appeal Brief (“Br.”) filed March 18,
2016; Examiner’s Answer (“Ans.”) mailed September 29, 2016; Final Office
Action (“Final Act.”) mailed October 19, 2015; and original Specification
(“Spec.”) filed December 10, 2009.
Appeal 2017-005466
Application 12/635,235
STATEMENT OF THE CASE
Appellants ’ Invention
Appellants’ invention relates to a system and a method that
regenerates wideband speech from narrowband speech, improves speech
naturalness, and alleviates metallic artefacts. Spec. 1:8—10, 2:15—16, 3:10-
11, 8:13—15, Title, and Abstract. According to Appellants, samples of a
narrowband speech signal are received and modulated, via a modulation
signal having a modulating frequency adapted to upshift each frequency in a
narrowband frequency band (associated with the narrowband speech signal)
by an amount determined by the modulating frequency. Abstract. The
narrowband frequency band that is to be translated upwards is selected to be
a band that is more likely to have a harmonic structure closer to that of a
missing wideband speech portion. Spec. 4:18—20, 8:13—18, 11:11—15. For
example, the narrowband frequency band can be selected to include
frequencies with a specified signal characteristic, e.g., good signal-to-noise
ratio, minimum echo, minimum distortion, and/or certain degree of voicing.
Spec. 5:5—10, 8:13—18, 9:25—30. The modulating frequency is similarly
selected so as to upshift the selected narrowband frequency band into a
target band of the missing wideband speech, thereby preserving harmonic
structure when regenerating the wideband speech. Spec. 4:18—20, 5:13—14,
6:20-24, 7:23—25, Abstract.
Claims 1 and 16 are independent. Representative claim 1 is
reproduced below with disputed limitations in italics'.
1. A method implemented in a receiver to regenerate
wideband speech from narrowband speech, the method
comprising:
2
Appeal 2017-005466
Application 12/635,235
receiving, by a decoder of the receiver, samples of a
narrowband speech signal in a first range of frequencies, the
narrowband speech signal missing a portion of wideband speech
from which the narrowband speech signal was generated;
evaluating signal characteristics of each of the
frequencies in the first range, an evaluation indicating which of
the frequencies in the first range when translated into a target
band are determined to be more likely, based on the signal
characteristics, to result in a reduced-artefact wideband speech
signal having a harmonic structure that is closer to a harmonic
structure of the portion of the wideband speech that is missing
than other frequencies in the first range;
modulating the received samples of the narrowband
speech signal with a modulation signal having a modulating
frequency adapted to upshift each frequency in the first range of
frequencies by the modulating frequency, the modulating
frequency selected to translate into the target band a selected
frequency band of the first range of frequencies that is selected
according to the evaluation;
filtering the modulated samples using a target band filter
to form a regenerated speech signal in the target band; and
combining the narrow band speech signal with the
regenerated speech signal in the target band to regenerate the
reduced-artefact wideband speech signal.
Br. 21—26 (Claims App’x).
Examiner’s Rejections & References
(1) Claims 1—5 and 7—15 stand rejected under 35 U.S.C. § 103(a)
as being unpatentable over Iser et al. (US 2008/0195392 Al; published
August 14, 2008; “Iser”) and Nilsson et al. (US 2003/0009327 Al;
published January 9, 2003; “Nilsson”). Final Act. 4—12.
(2) Claims 16—24 stand rejected under 35 U.S.C. § 103(a) as being
unpatentable over Iser, Nilsson, and Jax et al. (US 2003/0050786 Al;
published March 13, 2003; “Jax”). Final Act. 12—18.
3
Appeal 2017-005466
Application 12/635,235
Issues on Appeal
Based on Appellants’ arguments, the dispositive issue on appeal is
whether:
(1) the combination of Iser and Nilsson teaches or
suggests (i) “receiving, by a decoder of the receiver, samples of
a narrowband speech signal in a first range of frequencies, the
narrowband speech signal missing a portion of wideband speech
from which the narrowband speech signal was generated” and
(ii) “evaluating signal characteristics of each of the frequencies
in the first range, an evaluation indicating which of the
frequencies in the first range when translated into a target band
are determined to be more likely, based on the signal
characteristics, to result in a reduced-artefact wideband speech
signal having a harmonic structure that is closer to a harmonic
structure of the portion of the wideband speech that is missing
than other frequencies in the first range,” as recited in
Appellants’ independent claim 1. Br. 12—15; and
(2) the combination of Iser, Nilsson, and Jax teaches or
suggests “determining a signal characteristic for selecting which
of the frequencies in the first range of frequencies are to be
translated into the target band . . . the signal characteristic that is
determined being identifiable in the frequencies in the first range
that are more likely, when translated according to a pitch-
dependent spectral translation, to result in a regenerated
wideband speech signal having a harmonic structure that
approximates a harmonic structure of the portion of the wideband
speech that is missing,” as recited in Appellants’ independent
claim 16. Br. 16—19.
ANALYSIS
Claims 1—5 and 7—15
With respect to independent claim 1, the Examiner finds Iser’s
bandwidth extension system regenerates wideband speech from narrowband
speech using a spectral shifter for modulating a narrowband speech signal
4
Appeal 2017-005466
Application 12/635,235
(received acoustic signal) with a modulation signal having a modulating
frequency (of a predetermined shifting frequency value), the modulating
frequency adapted to upshift a frequency range of the received acoustic
signal “so that the shifted signal covers a frequency range suitable for
complementing the [received] acoustic signal.” Final Act. 5 (citing Iser
126). The Examiner further finds Iser filters modulated samples to form a
regenerated speech signal (upper bandwidth extension signal), and then
combines the narrowband speech signal (received acoustic signal) with the
regenerated speech signal, as claimed. Final Act. 5 (citing Iser || 30, 42).
To support the conclusion of obviousness, the Examiner relies on
Nilsson for teaching a narrowband speech signal missing a portion of
wideband speech from which the narrowband speech was generated, as
recited in claim 1. Final Act. 6 (citing Nilsson | 82). The Examiner also
finds Nilsson extracts at least one essential attribute from the narrowband
speech signal, thereby evaluating signal characteristics of each of the
frequencies in a first narrowband range, as claimed. Ans. 18 (citing Nilsson
| 82). The Examiner further finds Nilsson’s extracted essential attribute(s)
from the narrowband speech signal provide confidence levels and estimated
parameter values for wideband frequency components that are missing from
the narrowband speech signal, thereby teaching indicating which
narrowband frequencies, when translated, are determined to be more likely
to result in a reduced-artefact wideband speech signal, as claimed. Final
Act. 6—7 (citing Nilsson || 22, 83—84); Ans. 18—19.
Nilsson’s Figure 9, as described in paragraphs 81—84 of Nilsson, is
reproduced below with additional markings for illustration.
5
Appeal 2017-005466
Application 12/635,235
Nilsson’s Figure 9 illustrates a method of producing a wideband acoustic
signal on basis of a narrowband acoustic signal. Nilsson | 81.
Appellants dispute the Examiner’s factual findings regarding Nilsson
and Iser. Particularly, Appellants contend the combination of Nilsson and
Iser does not teach or suggest the claimed “receiving, by a decoder of the
receiver, samples of a narrowband speech signal in a first range of
frequencies, the narrowband speech signal missing a portion of wideband
speech from which the narrowband speech signal was generated.” Br. 12.
Appellants further argue “[t]here is simply no mention in Nilsson of
indicating which frequencies can be translated into a target frequency band
6
Appeal 2017-005466
Application 12/635,235
to produce an artefact-reduced wideband speech signal.” Br. 13. Rather,
Nilsson merely extracts an “essential attribute” from a narrowband acoustic
signal to obtain “parameters that estimate characteristics of a high band
speech signal,” which does not teach the claimed evaluation indicating
frequencies that are more likely, when translated, to produce an artefact-
reduced wideband signal. Br. 12—13 (citing Nilsson | 82).
Appellants also dispute the Examiner’s rationale for combining
Nilsson and Iser. Br. 13—15. Particularly, Appellants argue the Examiner’s
rationale for combining the references is “too general” and “does not address
why the skilled artisan .... would have selected particular components for
combination in the manner claimed.” Br. 14—15.
We do not find Appellants’ arguments persuasive. Instead, we find
the Examiner has provided a comprehensive response to Appellants’
arguments supported by a preponderance of evidence. Ans. 17—19. As
such, we adopt the Examiner’s findings and explanations provided therein.
Id. For additional emphasis, we note Appellants do not address the
Examiner’s rejection based on a combination of references, in which the
Examiner relied on Iser for “receiving, by a decoder of the receiver, samples
of a narrowband speech signal in a first range of frequencies,” and Nilsson
for a “narrowband speech signal missing a portion of wideband speech from
which the narrowband speech signal was generated.” Final Act. 4, 6 (citing
Iser 121; Nilsson || 22, 82 (“A first step 901 receives a segment of the
incoming narrow-band acoustic signal. . . . The wide-band acoustic signal
includes wide-band frequency components outside the spectrum of the
narrow-band acoustic signal”)); Ans. 18. Additionally, we agree with the
Examiner’s reasonable findings, not addressed by Appellants.
7
Appeal 2017-005466
Application 12/635,235
We further agree with the Examiner that Nilsson’s “extracting] at
least one essential attribute from the narrow-band acoustic signal” teaches
‘ ‘'evaluating signal characteristics of each of the frequencies'1'’ in the first
range as recited in Appellants’ claim 1. Ans. 18 (citing Nilsson | 82).
Nilsson’s “essential attribute” (parameter znb) “describes particular
properties of the received narrow-band acoustic signal aNB” and may include
“[t]he degree of voicing r, which represents one such essential feature.” See
Nilsson |47. Thus, Nilsson’s “essential attribute” represents a “signal
characteristic” of each of the frequencies in a “first range of frequencies,” as
claimed.
Further, Nilsson’s confidence level determines an allocation of signal
energies (estimated parameters) to frequency components, thereby indicating
frequencies that are more likely to appear in a wideband speech signal. See
Nilsson || 17, 64 (the confidence level “can thus also be used to control the
energy (or shape) of the bandwidth extended regions ... of the wide-band
acoustic signal,” such that “a relatively high energy is allocated to frequency
components being associated with a confidence level that represents a
comparatively high degree certainty” and “a relatively low energy is
allocated to frequency components if the confidence level being associated
with a confidence level that represents a comparatively low degree
certainty” (emphases added)), 68. Thus, Nilsson’s confidence levels and
estimated parameters (signal energies) indicate which frequencies are
determined to be more likely to result in a reduced-artefact wideband speech
signal, as recited in Appellants’ claim 1. Ans. 18—19; Final Act. 6—7.
We are also not persuaded by Appellants’ argument that “[t]here is
simply no mention in Nilsson of indicating which frequencies can be
8
Appeal 2017-005466
Application 12/635,235
translated into a target frequency band to produce an artefact-reduced
wideband speech signal” (see Br. 13 (emphasis added)). Particularly,
Nilsson discloses that a model of energy dependencies between narrowband
and wideband controls the signal energies allocated to regenerate a high
quality wideband speech signal from narrowband speech by spectral folding.
See Nilsson H 22, 35, 46, 74 (describing spectral folding), Fig. 3. Thus, we
agree with the Examiner that Nilsson teaches an evaluation indicating which
frequencies, when translated, are determined to be more likely to result in
reduced-artefact wideband speech, as claimed. Ans. 18—19; Final Act. 6—7.
Additionally, Iser discloses “shift[ing] the portion of the received acoustic
signal x(n) that is above a predetermined lower frequency value and/or
below a predetermined upper frequency value” so that “the shifted signal
covers a frequency range suitable for complementing the received acoustic
signal x(n),” also suggesting an evaluation indicating which frequency
translations are suitable for extending a narrowband speech signal x(n). See
Iser 1121,26-27.
As to Appellants’ argument that the Examiner’s combination of
Nilsson and Iser is improper (Br. 13—15), the Examiner has articulated
sufficient reasoning for refining Iser’s frequency upshifting technique for
speech extension, with Nilsson’s wideband speech extension based on
human voice characteristics, to improve the quality of reconstructed
speech—a concern shared by Iser and Nilsson. Final Act. 7 (citing Nilsson
122); see also Iser 1 6; Nilsson H 7, 13—14, 55. In particular, Iser and
Nilsson both teach the desirability of reducing artefacts in reconstructed
speech, which logically links the references together. See Iser 1 6; Nilsson
1113, 22. Enabling wideband speech reconstruction with reduced artefacts,
9
Appeal 2017-005466
Application 12/635,235
as taught by both references, would have been a desirable feature to a person
of ordinary skill, and that these teachings provide sufficient rationale to
combine the references.
In sum, there was “an apparent reason to combine the known elements
in the fashion claimed.” KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 417—
18 (2007). We find the Examiner provides “some articulated reasoning with
some rational underpinning to support the legal conclusion of obviousness.”
Id. at 418 (quoting In re Kahn, 441 F.3d 977, 988 (Fed. Cir. 2006)).
Thus, we agree with the Examiner the combination of Iser and Nilsson
teaches the “evaluating” step recited in Appellants’ claim 1.
Based on this record, we are not persuaded of Examiner error.
Accordingly, we sustain the Examiner’s obviousness rejection of
independent claim 1, and the rejection of its dependent claims 2—5 and 7—15
for which no substantive arguments are provided. Br. 15.
Claims 16—24
Independent claim 16 recites, inter alia:
means for controlling which frequencies in the first range of
frequencies are to be translated into a target band, the controlling
including determining a signal characteristic for selecting which
of the frequencies in the first range of frequencies are to be
translated into the target band, the signal characteristic
comprising one of a minimum echo, a minimum pre-processor
distortion, or a minimum degree of voicing, and the signal
characteristic that is determined being identifiable in the
frequencies in the first range that are more likely, when
translated according to a pitch-dependent spectral translation,
to result in a regenerated wideband speech signal having a
harmonic structure that approximates a harmonic structure of the
portion of the wideband speech that is missing.
10
Appeal 2017-005466
Application 12/635,235
Br. 24 (emphasis added). Appellants’ Specification further describes
determining a signal characteristic for selecting frequencies in a narrowband
range that are more likely, when translated according to a pitch-dependent
spectral translation, to result in regenerated wideband speech without
harmonic distortions (e.g., metallic-sounding artefacts). See Spec. 6:20—28,
7:20-8:3,8:13—18,9:19-10:2, 11:11—15. According to Appellants’
Specification:
[A] pitch dependent spectral translation translates a
frequency band (a range of frequencies from the narrowband
speech signal) into a target frequency band with properly
preserved harmonics. In the embodiment . . . the range of the
frequencies from 2-4kHz is translated to the target frequency
band of between 4 and 6kHz. However, it will be clear from the
following that these can be selected differently without diverging
from the concepts of the invention. . . .
The modulation frequency fmod is determined such as to
preserve the harmonic structure in the regenerated excitation
high band. In the present implementation, the modulating
frequency is normalised by the sampling frequency.
Taking the specific example, consider the pitch frequency
to be 180Hz, then the closest frequency to 2kHz that is an integer
multiple of the pitch frequency is floor(200/180)*180 (1980Hz).
Normalised by 1200Hz it becomes 0.165. For a sampling
frequency (after upsampling) of 12kHz and a value of 2kHz of
the frequency shift, the frequency fmod can be expressed as
fmod=floor(p/6)/p, where p represents the fractional pitch-lag.. . .
The frequency band of the narrow band speech x which is
translated can be selected to alleviate metallic artefacts by
selection of a frequency band that is more likely to have
harmonic structure closer to that of the missing (high) frequency
band by selection of a frequency band that includes frequencies
showing an identified signal characteristic. . . .
The signal characteristic can be chosen from a number of
different possibilities. . . . [A] possibility is that the block 30
11
Appeal 2017-005466
Application 12/635,235
determines the degree of voicing. According to one example, a
measure of the degree of voicing can be the normalised
correlation between the signal inside a frequency band and the
same signal one pitch-cycle earlier. Smoothed versions of this
measure can also be used to determine whether or not a
frequency should be included in the first range offrequencies for
translation.
Spec. 6:20-28, 7:20-8:3, 8:13—18, 9:19-10:2 (emphases added).
Appellants assert Iser, Nilsson, and Jax do not discuss a “pitch-
dependent spectral translation,” and do not teach or suggest a determined
signal characteristic that is “identifiable in the frequencies in the first range
that are more likely, when translated according to a pitch-dependent
spectral translation, to result in a regenerated wideband speech signal
having a harmonic structure that approximates a harmonic structure of the
portion of the wideband speech that is missing,” as recited in claim 16. Br.
16, 18 (emphases added). Thus, Appellants argue the combination of Iser,
Nilsson, and Jax does not teach or suggest all features of claim 16. Br. 16—
18.
We agree with Appellants. The Examiner has not responded to
Appellants’ arguments in the Answer. The Examiner, in the Final Action,
merely states that “Iser does not teach . . . the signal characteristic that is
determined being identifiable in the frequencies in the first range that are
more likely, when translated according to a pitch-dependent spectral
translation, to result in a regenerated wideband speech signal,” without
further addressing the claimed “pitch-dependent spectral translation.” Final
Act. 13—14 (emphasis added). We have also reviewed the cited portions of
Nilsson and Jax and do not find they disclose a signal characteristic
identifiable based on “a pitch-dependent spectral translation” of frequencies
12
Appeal 2017-005466
Application 12/635,235
in a range, as claimed. See Final Act. 14—16 (citing Nilsson || 22, 82—84;
Jax || 73—75, 80, 171). Absent evidence of such teachings or reasoning as
to why it would have been obvious to modify Iser, Nilsson, and Jax to
distinguish a signal characteristic for selecting narrowband frequencies
based on a pitch-dependent spectral translation, on the record before us, we
do not sustain the Examiner’s rejection of claim 16 and its dependent claims
17—24 under 35 U.S.C. § 103(a) over Iser, Nilsson, and Jax.
CONCLUSION
On the record before us, we conclude Appellants have demonstrated
the Examiner erred in rejecting claims 16—24, but not claims 1—5 and 7—15.
DECISION
As such, we AFFIRM the Examiner’s Final Rejection of claims 1—5
and 7—15. However, we REVERSE the Examiner’s Final Rejection of
claims 16—24.
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).
AFFIRMED-IN-PART
13