DETAILED ACTION
This communication is in response to the Amendments and Arguments filed on 07/09/2026.
Claims 1-11, 13-16, and 18-22 are pending and have been examined.
All previous objections/rejections not mentioned in this Office Action have been withdrawn by the examiner.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Arguments
Applicant's arguments filed 07/09/2026 have been fully considered but they are either not persuasive and/or are moot.
Regarding the 101 rejection, Applicant asserts on pgs 8-9 that a human cannot perform a training step and that the computation of an element-wise difference between two speech embedding vectors to compute a change value recite operations that describe an improvement to speech-based health-condition diagnosis and monitoring. The Examiner respectfully disagrees with these assertions. Training using specific data reads to a human looking at a specific set of data to learn rules and procedures for how to determine desired information from a set of provided input that has specific characteristics. As such, the claims are directed to an abstract idea. Additionally, it is not clear from the claim language how training the models with the data specified, or how the computation of a change value using element-wise differences results in a technological improvement to the ability to determine speech-based health-condition diagnosis and monitoring – only that it is a method for performing speech-based health-condition diagnosis and monitoring.
Applicant’s arguments regarding the 103 rejections have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Hence, Applicant’s arguments are either not persuasive or are moot.
Double Patenting
The nonstatutory double patenting rejection is based on a judicially created doctrine grounded in public policy (a policy reflected in the statute) so as to prevent the unjustified or improper timewise extension of the “right to exclude” granted by a patent and to prevent possible harassment by multiple assignees. A nonstatutory double patenting rejection is appropriate where the conflicting claims are not identical, but at least one examined application claim is not patentably distinct from the reference claim(s) because the examined application claim is either anticipated by, or would have been obvious over, the reference claim(s). See, e.g., In re Berg, 140 F.3d 1428, 46 USPQ2d 1226 (Fed. Cir. 1998); In re Goodman, 11 F.3d 1046, 29 USPQ2d 2010 (Fed. Cir. 1993); In re Longi, 759 F.2d 887, 225 USPQ 645 (Fed. Cir. 1985); In re Van Ornum, 686 F.2d 937, 214 USPQ 761 (CCPA 1982); In re Vogel, 422 F.2d 438, 164 USPQ 619 (CCPA 1970); In re Thorington, 418 F.2d 528, 163 USPQ 644 (CCPA 1969).
A timely filed terminal disclaimer in compliance with 37 CFR 1.321(c) or 1.321(d) may be used to overcome an actual or provisional rejection based on nonstatutory double patenting provided the reference application or patent either is shown to be commonly owned with the examined application, or claims an invention made as a result of activities undertaken within the scope of a joint research agreement. See MPEP § 717.02 for applications subject to examination under the first inventor to file provisions of the AIA as explained in MPEP § 2159. See MPEP § 2146 et seq. for applications not subject to examination under the first inventor to file provisions of the AIA . A terminal disclaimer must be signed in compliance with 37 CFR 1.321(b).
The filing of a terminal disclaimer by itself is not a complete reply to a nonstatutory double patenting (NSDP) rejection. A complete reply requires that the terminal disclaimer be accompanied by a reply requesting reconsideration of the prior Office action. Even where the NSDP rejection is provisional the reply must be complete. See MPEP § 804, subsection I.B.1. For a reply to a non-final Office action, see 37 CFR 1.111(a). For a reply to final Office action, see 37 CFR 1.113(c). A request for reconsideration while not provided for in 37 CFR 1.113(c) may be filed after final for consideration. See MPEP §§ 706.07(e) and 714.13.
The USPTO Internet website contains terminal disclaimer forms which may be used. Please visit www.uspto.gov/patent/patents-forms. The actual filing date of the application in which the form is filed determines what form (e.g., PTO/SB/25, PTO/SB/26, PTO/AIA /25, or PTO/AIA /26) should be used. A web-based eTerminal Disclaimer may be filled out completely online using web-screens. An eTerminal Disclaimer that meets all requirements is auto-processed and approved immediately upon submission. For more information about eTerminal Disclaimers, refer to www.uspto.gov/patents/apply/applying-online/eterminal-disclaimer.
Claims 1-5, 7-10, 13-16, and 18-22 are rejected on the ground of nonstatutory double patenting as being unpatentable over claims 1-19 of U.S. Patent No. 12125497. Although the claims at issue are not identical, they are not patentably distinct from each other because the claims of the issued patent/co-pending application anticipate the claims of the instant application. Please see below for the mapping in the table, where the bolded limitations indicate the corresponding limitations between the issued patent/co-pending application and instant application. With respect to the dependent claims, each of the claims map to a corresponding dependent claim of the issued patent/co-pending application or are found within the scope of the independent claim.
With respect to each of the dependent claims and independent claims, each claim corresponds numerically. Please see mapping that follows: Instant application claim (I), Issued Patent/Co-Pending App (P) - Claim 1 (I):Claim 1 (P), Claim 2 (I):Claim 1 (P), Claim 3 (I):Claim 2 (P), Claim 4 (I):Claim 1 (P), Claim 5 (I):Claim 3 (P), Claim 7 (I):Claim 6 (P), Claim 8 (I):Claim 7 (P), Claim 9 (I):Claim 8 (P), Claim 10 (I):Claim 8 (P), Claim 13 (I):Claim 12 (P), Claim 14 (I):Claim 13 (P), Claim 15 (I):Claim 14 (P), Claim 16 (I):Claim 15 (P), Claim 18 (I):Claim 17 (P), Claim 19 (I):Claim 15 (P), Claim 20 (I):Claim 19 (P), Claim 21 (I):Claim 1 (P), Claim 22 (I):Claim 1 (P).
Instant Application: 18796286
Issued Patent: 12125497
Claim 1: A computer-implemented method, comprising:
computing, with at least one model, a first speech embedding vector by processing a first audio signal corresponding to a first time period, wherein one or more of the at least one model is trained using a training corpus of audio files and training labels that indicate change values corresponding to the audio files;
computing, with the at least one model, a second speech embedding vector by processing a second audio signal corresponding to a second time period; and
computing, with the at least one model, a change value indicating a change in a health condition between the first time period and the second time period based at least in part on an element-wise difference between the first speech embedding vector and the second speech embedding vector.
Claim 1: A computer-implemented method, comprising:
training a neural network using a training corpus of audio files and training labels that indicate health condition labels corresponding to the audio files, change values corresponding to pairs of audio files, or a combination thereof;
receiving a first audio signal corresponding to a first time period, wherein the first audio signal comprises speech of a person;
computing a first feature vector from the first audio signal;
computing a first speech embedding vector by processing the first feature vector with the neural network;
obtaining a first health condition label indicating a health condition at the first time period;
combining the first health condition label with the first speech embedding vector using the neural network;
receiving a second audio signal corresponding to a second time period, wherein the second audio signal comprises a second speech of the person;
computing a second feature vector from the second audio signal;
computing a second speech embedding vector by processing the second feature vector with the neural network;
computing a difference vector comprising an elementwise difference between each element of the first speech embedding vector and each element of the second speech embedding vector;
computing a change value indicating a change in the health condition between the first time period and the second time period by processing the difference vector with a mathematical model; and
computing a second health condition label indicating the health condition at the second time period as a function of the first health condition label and the change value using the mathematical model.
System claim 9 of the instant application is rejected over system claim 8 of the issued patent/co-pending application using the same rationale as that provided in the table above for the method claims.
As to claim(s) 6 and 11, this/these claim(s) are rejected over the issued patent/co-pending application in view of Togawa et al. (U.S. Patent No. 11,721,357), hereinafter Togawa.
Please see the corresponding mappings below for further detail.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-11, 13-16, and 18-22 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding claim(s) 1, 9, and 16, the limitation(s) of computing, computing, and computing, as drafted, are processes that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, as well as mathematical calculations in prose. More specifically, the mental process of a human hearing a person’s first utterance, writing down features of the speech such as a word vector, using a mathematical algorithm to calculate another vector, repeating the process for a second utterance, writing out the differences between each corresponding element of the two final vectors, using another mathematical algorithm to calculate a value indicating the overall difference between the vectors that the human recognizes as indicative of a change in health. The recitation of “at least one model” in claims 1, 9, and 16, are comparable to a mathematical algorithm, which can be used by a human with pen and paper to perform the required calculations, and sets of learned rules for how to perform specific tasks, where the training of the model reads to the type of information a human uses to learn how to follow the rules. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, as well as mathematical calculations in prose, then it falls within the --Mental Processes—and –Mathematical Concepts-- groupings of abstract ideas. Accordingly, the claim(s) recite(s) an abstract idea.
This judicial exception is not integrated into a practical application because the recitation of “system” and “computer” in claim 9 and “non-transitory computer-readable media” and “processor” of claim 16 reads to generalized computer components, based upon the claim interpretation wherein the structure is interpreted using [0071-9] in the specification. Accordingly, these additional elements do not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The claim(s) is/are directed to an abstract idea.
The claim(s) do(es) not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract idea into a practical application, the additional element of using generalized computer components to perform multiple compute steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. The claim(s) is/are not patent eligible.
With respect to claim(s) 2, the claim(s) recite(s) “obtaining a first health condition label” and “computing a second health condition label”, which reads on a human reading a health condition value indicated by the final vector of the first utterance, and using the calculated difference value to identify a health condition value for the second utterance. No additional limitations are present.
With respect to claim(s) 3, the claim(s) recite(s) “computing the second health condition label”, which reads on a human adding the first health condition value to the difference value. No additional limitations are present.
With respect to claim(s) 4, 10, and 19, the claim(s) recite(s) computing and computing, which reads on a human writing down a transcription of a heard utterance and generating tokens for the individual words or sub-words, then repeating the process for a second utterance. No additional limitations are present.
With respect to claim(s) 5, 15, and 20, the claim(s) recite(s) the computing, which reads on a human using specific steps and information to perform process. No additional limitations are present.
With respect to claim(s) 6 and 11, the claim(s) recite(s) the at least one model comprises two different neural networks, which reads on two algorithms having certain mathematical features and/or two sets of learned rules. No additional limitations are present.
With respect to claim(s) 7, the claim(s) recite(s) details about what different health conditions may be identified, which reads on a human recognizing specific health conditions. No additional limitations are present.
With respect to claim(s) 8, 14, and 18, the claim(s) recite(s) “computing an anti-symmetric change value”, which reads on a human computing a specific type of difference between the vectors. No additional limitations are present.
With respect to claim(s) 13, the claim(s) recite(s) performing the limitations of claim 9 on a third audio signal, which reads on a human performing the same actions on a third utterance. No additional limitations are present.
With respect to claim(s) 21, the claim(s) recite(s) combining, which reads on a human writing different pieces of data together using a set of rules. No additional limitations are present.
With respect to claim(s) 22, the claim(s) recite(s) calculating the element-wise difference, which reads on a human performing specific mathematical calculations. No additional limitations are present.
These claims further do not remedy the judicial exception being integrated into a practical application and further fail to include additional elements that are sufficient to amount to significantly more than the judicial exception.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1-4, 6-11, 13, 14, 16, 18, 19, 21, and 22 is/are rejected under 35 U.S.C. 103 as being unpatentable over Provost et al. (U.S. PG Pub No. 2020/0075040), hereinafter Provost, in view of Togawa.
Regarding claim 1, 9, and 16, Provost teaches
(claim 1) A computer-implemented method ([0019]), comprising:
(claim 9) A system, comprising at least one computer configured to ((5:40-6:20),(23:52-61),(24:25-32)),:
(claim 16) One or more non-transitory computer-readable media comprising computer-executable instructions that, when executed, cause at least one processor to perform actions comprising ((5:40-6:20),(23:52-61),(24:25-32)):
computing, with at least one model, a first speech embedding vector by processing a first audio signal corresponding to a first time period (speech audio of a user captured over different sessions, i.e. a first audio signal corresponding to a first time period, where MFCCs are extracted from the audio data to determine a feature vector using different toolkits, i.e. at least one model, which is analyzed using a trained ML model, i.e. with at least one model, to determine an activation and valence value that is further stored, i.e. computing a first speech embedding vector [0040-4],[0048-51],[0054],[0070-2],[0095-6]), wherein one or more of the at least one model is trained using a training corpus of audio files and training labels that indicate change values corresponding to the audio files (the ML model may be trained to generate activation and valence values based on analyzing features pertaining to the outcomes that are of clinical concern, i.e. one or more of the at least one model is trained, where training data may be audio segments labeled with activation and valence labels, i.e. using a training corpus of audio files and training labels, where variations from valence and activation from the baseline can be associated with different mood states, i.e. indicate change values [0054-5],[0059-60],[0076],[0093],[0095-7]);
computing, with the at least one model, a second speech embedding vector by processing a second audio signal corresponding to a second time period (speech audio of a user captured over different sessions, i.e. a first audio signal corresponding to a first time period, where MFCCs are extracted from the audio data to determine a feature vector using different toolkits, i.e. at least one model, which is analyzed using a trained ML model, i.e. with at least one model, to determine an activation and valence value that is further stored, i.e. computing a first speech embedding vector [0040-4],[0048-50],[0054],[0070-2],[0095-6]); and
computing, with the at least one model, a change value indicating a change in a health condition between the first time period and the second time period …(data over call segments may be analyzed by a classifier or other model, i.e. computing with the at least one model, such as computing statistics over the acoustic features to determine a change in the user’s baseline indicative of a change in mental health, i.e. a change value indicating a change in a health condition between the first time period and the second time period [0028-9],[0031],[0038],[0040],[0043],[0063-8],[0076],[0093]).
While Provost provides determining a change in mental health based on a change in audio data from a subject baseline, Provost does not specifically teach an element-wise difference between two vectors, and thus does not teach
a change value indicating a change in a health condition between the first time period and the second time period based at least in part on an element-wise difference between the first speech embedding vector and the second speech embedding vector.
Togawa, however, teaches a change value indicating a change in a health condition between the first time period and the second time period based at least in part on an element-wise difference between the first speech embedding vector and the second speech embedding vector (a series of features amounts for a speaker, i.e. the first speech embedding vector and the second speech embedding vector, are calculated for each voice section, i.e. between the first time period and the second time period, where a determination unit obtains a change amount for each of the features, i.e. based at least in part on an element-wise difference between the first speech embedding vector and the second speech embedding vector, to identify a change in emotion from neutral (2:50-67),(3:50-4:2),(9:5-32),(12:21-61)).
Where Provost teaches that the change in emotion indicates a change in mental health [0028-9],[0031],[0038],[0040],[0043],[0063-8],[0076],[0093].
Provost and Togawa are analogous art because they are from a similar field of endeavor in processing audio to identify changes in user conditions. Thus, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to modify the determining a change in mental health based on a change in audio data from a subject baseline teachings of Provost with the comparison of feature amounts between two different sections as taught by Togawa. It would have been obvious to combine the references to enable accurate association of speakers with the voices, even when features change due to a change in emotion (Togawa (3:17-32)).
Regarding claim 2, Provost in view of Togawa teaches claim 1, and Togawa further teaches
obtaining a first health condition label indicating a health condition at the first time period (a determination unit determines the emotion of the speaker based on the voice signal included in the frame, where an initial frame, i.e. at the first time period, with the “first feature amount” is determined to be “neutral”, i.e. obtaining a first health condition label indicating a health condition (9:1-44)); and
computing a second health condition label indicating a health condition at the second time period by processing the first health condition label and the change value (a series of features amounts for a speaker are calculated for each voice section, i.e. at the second time period, where a determination unit obtains a change amount for each of the features compared to the first feature amount associated with a neutral frame, i.e. processing the first health condition label and the change value, to identify a the emotion for that frame, i.e. computing a second health condition label indicating a health condition (2:50-67),(3:50-4:2),(9:5-32),(12:21-61)).
Where Provost teaches that the change in emotion indicates a change in mental health [0028-9],[0031],[0038],[0040],[0043],[0063-8],[0076],[0093].
And where the motivation to combine is the same as previously presented.
Regarding claim 3, Provost in view of Togawa teaches claim 2, and Togawa further teaches
computing the second health condition label comprises adding the first health condition label and the change value (a series of features amounts for a speaker are calculated for each voice section, where a determination unit obtains a change amount for each of the features compared to the first feature amount associated with a neutral frame, i.e. adding the first health condition label and the change value, to identify a the emotion for that frame, i.e. computing the second health condition label (2:50-67),(3:50-4:2),(9:5-32),(12:21-61)).
Where Provost teaches that the change in emotion indicates a change in mental health [0028-9],[0031][0043],[0063-8],[0076],[0093].
And where the motivation to combine is the same as previously presented.
Regarding claims 4, 10, and 19, Provost in view of Togawa teaches claim 1, and Provost further teaches
computing the first speech embedding vector comprises processing a first feature vector with the at least one model, wherein the first feature vector is computed based on the first audio signal, and the first speech embedding vector is computed based on the first feature vector (speech audio of a user captured over different sessions, i.e. first audio signal, where MFCCs are extracted from the audio data to determine a feature vector using different toolkits, i.e. the first feature vector, which is analyzed using a trained ML model, i.e. processing a first feature vector with the at least one model, to determine an activation and valence value that is further stored, i.e. computing the first speech embedding vector…based on the first feature vector [0040-4],[0048-50],[0054],[0070-2],[0095-6]); and
computing the second speech embedding vector comprises processing a second feature vector with the at least one model, wherein the second feature vector is computed based on the second audio signal, and the second speech embedding vector is computed based on the second feature vector (speech audio of a user captured over different sessions, i.e. second audio signal, where MFCCs are extracted from the audio data to determine a feature vector using different toolkits, i.e. the second feature vector, which is analyzed using a trained ML model, i.e. processing a second feature vector with the at least one model, to determine an activation and valence value that is further stored, i.e. computing the second speech embedding vector…based on the second feature vector [0040-4],[0048-50],[0054],[0070-2],[0095-6]).
Regarding claims 6 and 11, Provost in view of Togawa teaches claims 4 and 10, and Provost further teaches
the at least one model comprises a first –model-- for computing the first feature vector and the second feature vector, and the at least one model comprises a second neural network for computing the first speech embedding vector and the second speech embedding vector (speech audio of a user captured over different sessions, where MFCCs are extracted from the audio data to determine a feature vector using different toolkits, i.e. a first –model-- for computing the first feature vector and the second feature vector, which is analyzed using a trained ML model, such as a CNN, to determine an activation and valence value that is further stored, i.e. a second neural network for computing the first speech embedding vector and the second speech embedding vector [0040-4],[0062-5],[0070-2],[0095-6]).
Togawa further teaches the at least one model comprises a first neural network for computing the first feature vector and the second feature vector (the calculation unit calculates the different feature amounts based on the voice signals included in each frame, i.e. computing the first feature vector and the second feature vector, where the calculations are performed by a neural network, i.e. at least one model comprises a first neural network (6:42-50),(8:62-9:3),(21:10-44)).
Where the motivation to combine is the same as previously presented.
Regarding claim 7, Provost in view of Togawa teaches claim 1, and Provost further teaches
the health condition corresponds to stress, depression, anxiety, post-traumatic stress disorder, concussion, Parkinson's disease, Alzheimer's disease, or congestive heart failure (the conditions associated with changes in emotion expression patterns include PTSD, bipolar disorder, anxiety, depression, and stress [0028-9],[0031],[0043],[0063-8],[0076],[0093]).
Regarding claims 8, 14, and 18, Provost in view of Togawa teaches claims 1, 9, and 16, and Togawa further teaches
computing an anti-symmetric change value (change vectors calculated based on a change between associated feature amounts, i.e. change value, may have different directions, such as a first feature amount changing in the positive direction going from neutral to sad, but changing in the negative direction going from neutral to angry (Fig. 18,(9:1-44),(12:21-61)).
Where the motivation to combine is the same as previously presented.
Regarding claim 13, Provost in view of Togawa teaches claim 9, and Provost further teaches
compute, with the at least one model, a third speech embedding vector by processing a third audio signal corresponding to a third time period (speech audio of a user captured over different sessions, i.e. a third audio signal corresponding to a third time period, where MFCCs are extracted from the audio data to determine a feature vector using different toolkits, i.e. at least one model, which is analyzed using a trained ML model, i.e. with at least one model, to determine an activation and valence value that is further stored, i.e. computing a third speech embedding vector [0040-4],[0048-51],[0054],[0070-2],[0095-6]); and
compute, with the at least one model, a second change value indicating a change in a health condition between the third time period and the second time period … (data over call segments may be analyzed by a classifier or other model, i.e. computing with the at least one model, such as computing statistics over the acoustic features to determine a change in the user’s baseline indicative of a change in mental health, i.e. a second change value indicating a change in a health condition between the third time period and the second time period [0028-9],[0031],[0038],[0040],[0043],[0063-8],[0076],[0093]).
Where Togawa further teaches a second change value indicating a change in a health condition between the third time period and the second time period based at least in part on a second element-wise difference between the third speech embedding vector and the second speech embedding vector (a series of features amounts for a speaker, i.e. the third speech embedding vector and the second speech embedding vector, are calculated for each voice section, i.e. between the third time period and the second time period, where a determination unit obtains a change amount for each of the features, i.e. a second change value…based at least in part on a second element-wise difference between the third speech embedding vector and the second speech embedding vector, to identify a change in emotion from neutral (2:50-67),(3:50-4:2),(9:5-32),(12:21-61)).
Where Provost teaches that the change in emotion indicates a change in mental health [0028-9],[0031],[0038],[0040],[0043],[0063-8],[0076],[0093].
And where the motivation to combine is the same as previously presented.
Regarding claim 21, Provost in view of Togawa teaches claim 1, and Provost further teaches
combining, with the at least one model, the first speech embedding vector with a first health condition label indicating a health condition at the first time period (speech audio of a user captured over different sessions, i.e. first time period, where MFCCs are extracted from the audio data to determine a feature vector using different toolkits, i.e. at least one model, which is analyzed using a trained ML model, i.e. with at least one model, to determine an activation and valence value that is further stored, i.e. computing a first speech embedding vector, and also assigned mood labels, such as by classifiers, i.e. combining with the at least one model the first speech embedding vector with a first health condition label indicating a health condition [0031],[0037],[0040-4],[0048-51],[0054],[0063-6],[0070-2],[0095-6]).
Regarding claim 22, Provost in view of Togawa teaches claim 1, and Togawa further teaches
calculating the element-wise difference by calculating a difference between each element of the first speech embedding vector and each element of the second speech embedding vector (a series of features amounts for a speaker, i.e. the first speech embedding vector and the second speech embedding vector, are calculated for each voice section, where a determination unit obtains a change amount for each of the features, i.e. calculating the element-wise difference by calculating a difference between each element, to identify a change in emotion from neutral (2:50-67),(3:50-4:2),(9:5-32),(12:21-61)).
Where the motivation to combine is the same as previously presented.
Claim(s) 5, 15, and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Provost, in view of Togawa, and further in view of Pappagari et al. (“Using state of the art speaker recognition and natural language processing technologies to detect Alzheimer’s disease and assess its severity”, ISCA 2020), hereinafter Pappagari.
Regarding claims 5, 15, and 20, Provost in view of Togawa teaches claims 4, 10, and 19, and Provost further teaches
computing the first feature vector comprises (i) performing speech recognition on the first audio signal to obtain recognized text …(extracted speech and language features, i.e. computing the first feature vector, includes generating transcripts using ASR tools, i.e. performing speech recognition on the first audio signal to obtain recognized text [0062-66]).
While Provost in view of Togawa provides determining features from transcripts, Provost in view of Togawa does not specifically teach obtaining word-piece encodings, and thus does not teach
computing the first feature vector comprises … (ii) obtaining word-piece encodings corresponding to the recognized text.
Pappagari, however, teaches computing the first feature vector comprises … (ii) obtaining word-piece encodings corresponding to the recognized text (inputs to a BERT model are tokens from a spoken transcript, i.e. recognized text, where the transcript has been tokenized into sub-words using a WordPiece tokenizer, i.e. obtaining word-piece encodings corresponding to the recognized text (Sec. 3.2), where features extracted for further processing include both MFCCs, and sub-word tokens from a transcript, i.e. computing the first feature vector, such as a transcript obtained using automatic speech recognition (Sec. 3.1.1),(Sec. 3.2),(Sec. 5)).
Provost, Togawa, and Pappagari are analogous art because they are from a similar field of endeavor in identifying health and emotions from the analysis of speech. Thus, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to modify the determining features from transcripts teachings of Provost, as modified by Togawa, with the use of WorkPiece tokenization as taught by Pappagari. It would have been obvious to combine the references to improve detection and assessment results by fusing acoustic and text modalities (Pappagari Abstract).
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NICOLE A K SCHMIEDER whose telephone number is (571)270-1474. The examiner can normally be reached 8:00 - 5:00 M-F.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Pierre-Louis Desir can be reached at (571) 272-7799. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/NICOLE A K SCHMIEDER/Primary Examiner, Art Unit 2659
