Telefonaktiebolaget LM Ericsson (publ)

Patent Trials and Appeals Board

Jan 10, 2022

2021000301 (P.T.A.B. Jan. 10, 2022)

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. 15/969,139 05/02/2018 Martin Sehlstedt 4015-10390 / P32438-US4 5816 24112 7590 01/10/2022 COATS & BENNETT, PLLC 1400 Crescent Green, Suite 300 Cary, NC 27518 EXAMINER SHARMA, NEERAJ ART UNIT PAPER NUMBER 2659 MAIL DATE DELIVERY MODE 01/10/2022 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 MARTIN SEHLSTEDT Appeal 2021-000301 Application 15/969,139 Technology Center 2600 Before BRADLEY W. BAUMEISTER, MICHAEL J. STRAUSS, and AMBER L. HAGY, Administrative Patent Judges. HAGY, Administrative Patent Judge. DECISION ON APPEAL STATEMENT OF THE CASE Pursuant to 35 U.S.C. § 134(a), Appellant1 appeals from the Examiner’s decision to reject claims 1-8, 10-17, and 19.2 We have jurisdiction under 35 U.S.C. § 6(b). We reverse. 1 “Appellant” herein refers to “applicant” as defined in 37 C.F.R. § 1.42(a). Appellant identifies the real party in interest as Telefonaktiebolaget LM Ericsson. Appeal Br. 2. 2 Claims 1-19 are pending; claims 9 and 18 are objected to as dependent on rejected base claims but otherwise deemed allowable. See Final Act. 1, 10. Appeal 2021-000301 Application 15/969,139 2 CLAIMED SUBJECT MATTER Appellant’s disclosure “relates to a method and a voice activity detector and in particular to an improved voice activity detector for handling e.g. non stationary background noise.” Spec. ¶ 2. By way of background, Appellant’s Specification describes that encoding of speech may only require the encoder to be actively encoding when a speaker is actually talking, and “the rest can be encoded using comfort noise.” Id. ¶ 3. To allow such selective encoding, “it is important to detect the periods of speech in the input signal.” Id. ¶ 4. This detection may be done by utilizing Voice Activity Detectors (VADs), which detect when a person is speaking. Id. ¶ 4. A VAD typically produces a VAD decision as output, which is a decision as to whether a given frame contains speech or noise. Id. This VAD decision is traditionally used for adding hangover, which may adapt the VAD decision to form a final decision. Id. ¶¶ 4, 6. The addition of hangover is typically to mitigate the risk of improperly clipping the speech off (e.g., clipping the tail end of speech, during brief pauses, or during bursty speech patterns). Id. ¶ 6. Appellant’s Specification describes “conventional approaches” that are used to improve voice detection by combining multiple detection factors and making an activity decision based on those combined factors. Id. ¶ 8. Appellant’s Specification contrasts such conventional approaches with the approach described and claimed in Appellant’s application, in which multiple decision signals are generated and then combined to generate a combined decision signal as input to a hangover addition circuit. See id. ¶ 19. Appellant’s Specification explains that, “by allowing multiple VAD’s to work in parallel and then combine the outputs, it is possible to exploit the Appeal 2021-000301 Application 15/969,139 3 strengths from the different VAD’s without suffering too much from each VAD’s limitations.” Id. ¶ 16. Claims 1, 10, and 19 are independent. Claim 1, reproduced below, illustrates the claimed subject matter: 1. A method, implemented in a signal activity detector (SAD), for detecting activity in an input signal, the method comprising: combining at least three decision signals to generate a combined decision signal as input to a hangover addition circuit of the SAD, each of the decision signals indicating whether or not activity is detected in the input signal according to respective decision criteria; sending the combined decision signal to the hangover addition circuit to generate a final decision signal of the SAD as to whether or not activity is detected in the input signal. Appeal Br. 16 (Claims App.). REFERENCES The prior art relied upon by the Examiner is: Name3 Reference Date Marchok US 6,522,746 B1 Feb. 18, 2003 Wang US 2009/0089053 A1 April 2, 2009 3 All prior art references are identified by the first-named inventor only. Appeal 2021-000301 Application 15/969,139 4 REJECTIONS4 Claims 1-4, 8, 10-13, 17, and 19 stand rejected under 35 U.S.C. § 102(b) as anticipated by Marchok. Final Act. 4-7. Claims 5-7 and 14-16 stand rejected under 35 U.S.C. § 103(a) as unpatentable over the combined teachings of Marchok and Wang. Final Act. 8-10. OPINION The Examiner rejects claims 1-4, 8, 10-13, 17, and 19 as anticipated by Marchok. Final Act. 4-7. For essentially the reasons argued by Appellant (Appeal Br. 7-9; Reply Br. 2-4), we are persuaded of Examiner error at least in the finding that Marchok discloses “combining at least three decision signals to generate a combined decision signal as input to a hangover addition circuit of the SAD, each of the decision signals indicating whether or not activity is detected in the input signal according to respective decision criteria,” as recited in claim 1 and commensurately recited in independent claims 10 and 19.5 See Final Act. 5. With regard to the step of combining three decision signals, the Examiner finds Marchok discloses “the combining of individual VAD 4 The Leahy-Smith America Invents Act (“AIA”) included revisions to 35 U.S.C., including sections 102 and 103, that became effective on March 16, 2013. Because the present application claims priority to an application filed before March 16, 2013, the Examiner applies the pre-AIA version of the statutory basis for unpatentability. See Final Act. 2. 5 Appellant’s contentions present additional issues. Because the identified issue is dispositive of Appellant’s arguments on appeal, we do not reach the additional issues. Appeal 2021-000301 Application 15/969,139 5 decisions in the speech encoder.” Final Act. 3. In particular, the Examiner finds Figure 2 of Marchok depicts “the use of the combined decision of three VADs, each contained in an echo canceller, noise reduction and auto-level control circuit. The combined decision of these three VADs is input to the speech coder circuit, which acts as the hangover addition circuit.” Final Act. 5 (citing Marchok Fig. 2, 5:14-32). Appellant argues the Examiner’s findings are in error because Marchok does not disclose combining decision signals, but instead discloses serially processing a voice signal using blocks that contain VADs: As illustrated in Figure 2, the sole input of each subsystem is a voice signal, and the sole output of each subsystem is the voice signal after having been processed. The decision signals produced by the VADs of the respective subsystems are not exposed to each other, much less combined in any way into a “combined decision signal” as claimed. . . . Marchok teaches decision signals that are only taught to activate respective signal processing on a voice signal in series, and none of which are taught to be combined with any other. Appeal Br. 8. The Examiner responds by stating that “each individual VAD in Marchok does and will carry out voice detection, which is indeed the purpose of a VAD.” Ans. 3. The Examiner further finds: [T]he action of a noise reduction box cannot be carried out without the actions of the VAD of the previous box, i.e. echo canceller VAD. This is true of all the VADs in this chain as the signal output of one is the signal input of the other. Thus, the echo canceller box “as a whole” acts as a signal activity detector (SAD) with its own VAD. Further, the voice detection is not the sole output of the echo canceller VAD. The Echo canceller VAD is combined with an echo estimator to produce a joint output signal. Similarly, the output of the noise reduction VAD is combined with a noise estimator and the Appeal 2021-000301 Application 15/969,139 6 output of the auto level control VAD is combined with a signal estimator. In conclusion, the Examiner argues that it is the combination of the Echo Canceller VAD output with the output of the Echo Estimator that generates a joint/combined output signal and it is this joint/combined signal that is detected by the Echo Canceller block. This functionality is repeated for each of the other blocks, i.e. noise reduction block and the auto-level controller block. Id. at 4-5. As Appellant aptly contends, the Examiner essentially finds Marchok discloses the claimed “combined decision signal” because it discloses processing blocks that work in series wherein each processing block includes a VAD. See Reply Br. 2. Appellant acknowledges that Marchok discloses processing blocks that build upon each other, but argues that this fails to disclose a combined decision signal, as claimed. Reply Br. 2-4. Appellant explains, as follows: Marchok clearly teaches that each processing block performs the corresponding operation on the input voice signal, and does not output any type of activity decision. See at least col. 4, ll. 26-38, which discusses how the voice processing operations (i.e., echo cancellation, noise reduction, ALC, speech coding, etc.) are sequentially performed on the input voice signal 102, as shown in Figure 2. As such, each processing block receives an input voice signal and outputs a processed version (e.g., without echo, with reduced noise, etc.) of that voice signal. The independent claims explicitly recite that each decision signal indicates whether activity is detected in the input signal according to respective decision criteria. Nothing in Marchok teaches, or even implies that each processing block output (or any of the processing block outputs) is a decision indicating whether activity is present in the input signal. Thus, the output of each processing block in Figure 2 of Marchok does not act as the claimed signal activity detector as asserted by the Examiner. Appeal 2021-000301 Application 15/969,139 7 The Appellant also notes that the independent claims explicitly recite that the three (or more) decision signals (each indicating whether or not activity is detected in the input signal according to respective decision criteria) are combined to generate a combined decision signal as input to a hangover addition circuit of the SAD. Marchok clearly does not provide any such teaching. The fact that the VAD from one block, e.g., the EC block, influences processing applied within that block, and thus influences the input to the next block, e.g., the NR block, demonstrates a potential interaction between the blocks. Such interactions, however, neither represent a combination of the VAD outputs, nor represent a combination of decision signals, nor indicate whether or not activity is detected in the input signal, nor generate a combined decision signal that is subsequently sent to a hangover addition circuit to generate a final decision of the SAD as to whether activity is detected in an input signal. Id. at 2-3. We are persuaded of Examiner error in the rejection. As Appellant contends, and we agree, Marchok discloses an interaction between blocks that necessarily occurs because the output of one block is serially input to the next block. See Reply Br. 4. This demonstrates how the VAD from one block influences the signal output by the corresponding processing block, and thus influences the input signal to the next processing block. See id. But, as Appellant further contends, and we agree, this disclosure does not demonstrate a combination of decision signals that is then used to generate a combined decision signal. See id. Simply put, the cited disclosures of Marchok do not support the Examiner’s findings that Marchok discloses combining at least three decision signals to generate a combined decision signal. We are, therefore, persuaded that the Examiner’s findings are insufficient to show that Marchok discloses at least “combining at least three decision signals to Appeal 2021-000301 Application 15/969,139 8 generate a combined decision signal as input to a hangover addition circuit of the SAD, each of the decision signals indicating whether or not activity is detected in the input signal according to respective decision criteria,” as recited in independent claim 1, and commensurately recited in independent claims 10 and 19. For the foregoing reasons, we are persuaded of Examiner error in the 35 U.S.C. § 102(a)(2) rejection of independent claims 1, 10, and 19, and we, therefore, do not sustain that rejection. The dependent claims stand with their respective independent claims.6 CONCLUSION The Examiner’s decision rejecting claims 1-8, 10-17, and 19 is reversed. DECISION SUMMARY In summary: Claims Rejected 35 U.S.C. § Reference(s)/Basis Affirmed Reversed 1-4, 8, 10- 13, 17, 19 102(b) Marchok 1-4, 8, 10- 13, 17, 19 5-7, 14-16 103(a) Marchok, Wang 5-7, 14-16 Overall Outcome 1-8, 10-17, 19 REVERSED 6 Although the Examiner rejects dependent claims 5-7 and 14-16 under a different ground (under 35 U.S.C. § 103(a) over the combined teachings of Marchok and Wang), the Examiner does not rely on Wang in a manner that cures the deficiencies in the Examiner’s findings regarding Marchok. See Final Act. 8-10.