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 23 March have been fully considered but they are not persuasive.
Applicant has amended independent claims 1, 15, and 20 to include additional limitations and simply argues that Huffman does not teach or suggest the added limitations. Applicant argues that all 103 rejections should be withdrawn for the same reasons.
However, it is noted that the limitation “when the indication is that policy- violating content is present in the audio of the chunk, the detection model further outputs a time-span within the chunk that corresponds to the policy-violating content” was previously recited in dependent claim 10, now canceled. As noted in the prior Office Action, Huffman discloses this limitation.
Similarly, the limitations “wherein the policy-violating content is a word, wherein the portion of the audio of the chunk is a portion within the time-span that corresponds to a last syllable of the word” and “wherein the delay is greater than or equal to a sum of the portion within the time-span and an inference time of the detection model for the chunk” were previously recited in dependent claim 12, now cancelled. As noted in the prior Office Action, Huffman discloses the first limitation. For the second limitation, Huffman discloses a delay is greater than or equal to a sum of the time span and an inference time of the detection model for the chunk. Since a portion of the time span will be less than the entire time span, then the delay will inherently also be greater than or equal to a sum of the portion within the time span and the inference time of the detection model for the chunk.
Thus, the only subject matter not addressed in the prior Office Action is the limitations “wherein dividing the real-time audio stream into the plurality of chunks comprises dynamically adjusting chunk sizes based on speech dynamics and environmental noise levels” and “wherein replacing the portion of the audio of the chunk with different audio includes overlaying the detected policy-violating content with alternative audio”.
Huffman discloses dividing the real-time audio stream into variably sized chunks (paragraph [0054]), but does not disclose the chunk sizes are based on speech dynamics and environmental noise levels. Wang et al. (cited below) disclose an audio signal segmentation algorithm that dynamically adjusts chunk sizes based on speech dynamics and environmental noise levels. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to dynamically adjust chunk sizes in this manner in view of Wang et al. for the reasons provided below.
Huffman discloses muting at least a portion of an audio chunk and does not disclose replacing a portion of the chunk with different audio. However, Pearce et al. (previously cited in the rejection of claim 12) discloses replacing a portion of an audio chunk corresponding to a last syllable of a word by overlaying the detected policy-violating content with alternative audio. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to overlay the detected policy-violating content with alternative audio in view of Pearce et al. for the reasons provided below.
For the reasons given above, claims 1, 15, and 20 are rejected under 35 U.S.C. 103 below. Applicant’s amendments necessitated the new grounds of rejection. Accordingly, this action is FINAL.
Claim Rejections - 35 USC § 103
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) 1-2, 4, 6-9, 13-16, 18 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Huffman et al. (U.S. Patent Application Pub. No. 2023/0321546, hereinafter “Huffman”), in view of Wang et al. (U.S. Patent Application Pub. No. 2007/0271093, hereinafter “Wang”), and further in view of Pearce et al. (U.S. Patent Application Pub. No. 2022/0059071, hereinafter “Pearce”).
In regard to claim 1, Huffman discloses a computer-implemented method comprising:
obtaining a real-time audio stream of voice chat communication (audio within a real-time voice chat is obtained, paragraph [0087]);
dividing the real-time audio stream into a plurality of chunks, wherein each chunk comprises audio from the real-time audio stream from a respective time window (the input audio utilizes an input buffer to predict whether the buffer contains portions of content to be redacted, paragraph [0088]);
for each chunk,
providing the audio of the chunk to a machine-learning based detection model (the buffered audio is provided to a prediction model that predicts whether to-be-redacted content was spoken, paragraphs [0089-0091]);
obtaining, as output of the detection model, an indication whether policy-violating content is present in the audio of the chunk, wherein the detection model comprises an encoder that encodes the audio of the chunk into a feature vector and a classifier that generates the indication of whether policy-violating content is present in the audio of the chunk based on the feature vector (the first stage of the prediction model encodes the audio into an ordered sequence of distributions over phoneme probabilities parameterized by model weight vectors, paragraph [0089]; the sequence of phoneme distributions is compared to a lexicon of to-be-redacted words to determine whether to-be-redacted words were spoken or soon-to-be spoken in the audio of the input buffer, paragraphs [0090-0091]), and wherein, when the indication is that policy-violating content is present in the audio of the chunk, the detection model further outputs a time-span within the chunk that corresponds to the policy-violating content, and wherein the portion of the audio of the chunk is from within the time-span (see Fig. 1B, a time period t2-t3 is determined to include a target word to be redacted, paragraph [0112]);
in response to the indication that policy-violating content is present in the audio of the chunk, modifying the audio of the chunk to at least one of: mute at least a portion of the audio of the chunk or replace the at least a portion of the audio of the chunk with different audio, wherein the policy-violating content is a word (when the audio content is determined to likely include to-be-redacted words, the audio samples are set to zero before being copied to an output buffer, paragraph [0091]) ; and
providing the audio stream having the chunk with the modified audio to a client device for playback at the client device, wherein the audio stream is provided with a delay (the redacted audio stream is provided to an operator of a computer, Fig. 1A, 121, 122, 123, paragraph [0100]; with an introduced latency, paragraphs [0102-0103]), wherein the delay is greater than or equal a sum of the portion within the time-span and an inference time of the detection model for the chunk (a determination is made whether the full duration of a to-be-redacted target word has been spoken and the introduced latency is greater than the audio interval, paragraphs [0091] and [0103]; this will inherently be greater than or equal to the duration of a portion of the whole word and the introduced latency).
Huffman does not disclose dividing the real-time audio stream into the plurality of chunks comprises dynamically adjusting chunk sizes based on speech dynamics and environmental noise levels.
Wang discloses a method for dividing a real-time audio stream into a plurality of chunks that comprises dynamically adjusting chunk sizes based on speech dynamics and environmental noise levels (speech segments are determined by performing audio activity detection and determining a crossing rate indicative of speech, paragraph [0027]; the audio activity detection using environmental noise levels to first identify frames of noisy audio, and then merging frames into variable sized segments, paragraphs [0029-0031]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to divide the real-time audio stream into dynamically adjusted chunk sizes based on speech dynamics and environmental noise levels, because it would allow the audio stream to be segmented in practical noisy environments with low SNR, as taught by Wang (paragraph [0044]).
Huffman and Wang do not disclose the portion of the audio of the chunk is a portion within the time-span that corresponds to a last syllable of the word, and wherein replacing the portion of the audio of the chunk with different audio includes overlaying the detected policy-violating content with alternative audio.
Pearce discloses a method for replacing policy-violating content in an audio signal, wherein a portion of an audio chunk is a portion within the time-span that corresponds to a last syllable of the word, and wherein replacing the portion of the audio of the chunk with different audio includes overlaying the detected policy-violating content with alternative audio (a sound replacement algorithm replaces the last syllable of a policy-violating word with alternative audio, paragraph [0033]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to replace the last syllable of the word by overlaying the detected policy-violating content with alternative audio, because modifying only the last syllable of the word would allow the original audio to be modified with a reduced delay, as taught by Pearce (paragraph [0034]).
In regard to claim 2, Huffman discloses the feature vector represents speech characteristics of the audio of the chunk, and wherein the encoder comprises one or more convolutional layers that extract local features from the audio of the chunk (the input audio is processed using a plurality of convolutional layers to produce probabilities of redactions, paragraphs [0113-0115]).
In regard to claim 4, Huffman discloses performing pre-processing that includes one or more of:
removing background noise from the audio stream;
converting the audio stream into a particular digital format;
extracting Mel-Frequency Cepstral Coefficients (MFCCs) from the audio stream; and
combinations thereof (the audio stream is converted to a specific format, including MFCC representations of the audio stream, paragraph [0087]).
In regard to claim 6, Huffman discloses the classifier generates the indication as one of:
policy-violating content is present in the audio of the chunk or policy-violating content is absent from the audio of the chunk (the prediction model predicts whether target speech to be redacted is present or non-target speech that is not to be redacted, paragraphs [0116-0117]).
In regard to claim 7, Huffman discloses the classifier generates the indication as a category for the audio of the chunk, and wherein the category is one of: non-policy violating content or policy-violating content that is one of: hate speech, swearing, spam, or harassment (target speech comprises profanity, epithets, insults, etc., paragraph [0119]).
In regard to claim 8, Huffman discloses modifying the audio of the chunk to mute or replace the policy-violating content is based on the category (speech classified as target speech to be redacted is removed, paragraphs [0116-0117]; the target speech comprising the category of policy-violating content, paragraph [0119]).
In regard to claim 9, Huffman discloses the detection model is trained to detect whether an input audio chunk includes one or more words from a predefined vocabulary, and wherein the indication that policy-violating content is present in the audio of the chunk is generated when the chunk is determined to include at least one word from the predefined vocabulary (target speech terms that are to be redacted, paragraph [0117]).
In regard to claim 13, Huffman discloses the encoder is a pre-trained encoder from an automatic speech recognition (ASR) model that includes the encoder and a separate decoder that is trained to decode the feature vector into text (an encoder extracts a sequence of phoneme distributions, which is further processed by a decoder language model to produce a set of candidate sequences of likely spoken words, where the model is distilled from a larger model representing all speech in a given language to only entries which are relevant to the target words to-be-redacted, paragraphs [0089-0091]).
In regard to claim 14, Huffman discloses the machine-learning based detection model is trained by:
obtaining a training set (a set of training data, paragraph [0083]), wherein each element of the training set comprises:
a training chunk of audio from a respective time window of a real-time audio stream (the set of data comprises examples from a voice chat moderation system, paragraph [0083]); and
a training label indicative of a presence of policy-violating content in the chunk of audio (the set of data comprises labels indicating target data that should be redacted, paragraph [0083]); and
training the machine-learning based detection model via supervised learning (the training procedure is performed, paragraph [0083]), wherein the training comprises, for each element in the training set:
obtaining, by application of the machine-learning based detection model to the training chunk of audio, an indication of whether policy-violating content is present in the training chunk of audio, wherein the machine-learning based detection model comprises an encoder that encodes the training chunk of audio into a training feature vector and a classifier that generates the indication of whether policy-violating content is present in the training chunk of audio based on the training feature vector (the prediction model produces a candidate prediction of whether the audio contains target content that is to be redacted, paragraph [0083]; by producing phoneme probabilities and determining whether a target word that is to be redacted has been spoken, paragraphs [0089-0091]);
determining a loss value based on a comparison of the indication of whether policy-violating content is present in the training chunk of audio and the training label (the predictions are compared with the labels to determine the accuracy of the prediction, paragraph [0083]); and
modifying one or more parameters of the machine learning-based detection model based on the loss value (based on the comparison, the parameters of the prediction model are updated, paragraph [0083]).
In regard to claim 15, Huffman discloses a system (Fig. 2, 200) comprising:
one or more processors (processor 230); and
memory coupled to the one or more processors with instructions stored thereon that, when executed by the one or more processors, cause the one or more processors to perform or control performance of operations (memory 220) comprising:
obtaining a real-time audio stream of voice chat communication (audio within a real-time voice chat is obtained, paragraph [0087]);
dividing the real-time audio stream into a plurality of chunks, wherein each chunk comprises audio from the real-time audio stream from a respective time window (the input audio utilizes an input buffer to predict whether the buffer contains portions of content to be redacted, paragraph [0088]);
for each chunk,
providing the audio of the chunk to a machine-learning based detection model (the buffered audio is provided to a prediction model that predicts whether to-be-redacted content was spoken, paragraphs [0089-0091]);
obtaining, as output of the detection model, an indication whether policy-violating content is present in the audio of the chunk, wherein the detection model comprises an encoder that encodes the audio of the chunk into a feature vector and a classifier that generates the indication of whether policy-violating content is present in the audio of the chunk based on the feature vector (the first stage of the prediction model encodes the audio into an ordered sequence of distributions over phoneme probabilities parameterized by model weight vectors, paragraph [0089]; the sequence of phoneme distributions is compared to a lexicon of to-be-redacted words to determine whether to-be-redacted words were spoken or soon-to-be spoken in the audio of the input buffer, paragraphs [0090-0091]), and wherein, when the indication is that policy-violating content is present in the audio of the chunk, the detection model further outputs a time-span within the chunk that corresponds to the policy-violating content, and wherein the portion of the audio of the chunk is from within the time-span (see Fig. 1B, a time period t2-t3 is determined to include a target word to be redacted, paragraph [0112]);
in response to the indication that policy-violating content is present in the audio of the chunk, modifying the audio of the chunk to at least one of: mute at least a portion of the audio of the chunk or replace the at least a portion of the audio of the chunk with different audio, wherein the policy-violating content is a word (when the audio content is determined to likely include to-be-redacted words, the audio samples are set to zero before being copied to an output buffer, paragraph [0091]) ; and
providing the audio stream having the chunk with the modified audio to a client device for playback at the client device, wherein the audio stream is provided with a delay (the redacted audio stream is provided to an operator of a computer, Fig. 1A, 121, 122, 123, paragraph [0100]; with an introduced latency, paragraphs [0102-0103]), wherein the delay is greater than or equal a sum of the portion within the time-span and an inference time of the detection model for the chunk (a determination is made whether the full duration of a to-be-redacted target word has been spoken and the introduced latency is greater than the audio interval, paragraphs [0091] and [0103]; this will inherently be greater than or equal to the duration of a portion of the whole word and the introduced latency).
Huffman does not disclose dividing the real-time audio stream into the plurality of chunks comprises dynamically adjusting chunk sizes based on speech dynamics and environmental noise levels.
Wang discloses a method for dividing a real-time audio stream into a plurality of chunks that comprises dynamically adjusting chunk sizes based on speech dynamics and environmental noise levels (speech segments are determined by performing audio activity detection and determining a crossing rate indicative of speech, paragraph [0027]; the audio activity detection using environmental noise levels to first identify frames of noisy audio, and then merging frames into variable sized segments, paragraphs [0029-0031]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to divide the real-time audio stream into dynamically adjusted chunk sizes based on speech dynamics and environmental noise levels, because it would allow the audio stream to be segmented in practical noisy environments with low SNR, as taught by Wang (paragraph [0044]).
Huffman and Wang do not disclose the portion of the audio of the chunk is a portion within the time-span that corresponds to a last syllable of the word, and wherein replacing the portion of the audio of the chunk with different audio includes overlaying the detected policy-violating content with alternative audio.
Pearce discloses a method for replacing policy-violating content in an audio signal, wherein a portion of an audio chunk is a portion within the time-span that corresponds to a last syllable of the word, and wherein replacing the portion of the audio of the chunk with different audio includes overlaying the detected policy-violating content with alternative audio (a sound replacement algorithm replaces the last syllable of a policy-violating word with alternative audio, paragraph [0033]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to replace the last syllable of the word by overlaying the detected policy-violating content with alternative audio, because modifying only the last syllable of the word would allow the original audio to be modified with a reduced delay, as taught by Pearce (paragraph [0034]).
In regard to claim 16, Huffman discloses the feature vector represents speech characteristics of the audio of the chunk, and wherein the encoder comprises one or more convolutional layers that extract local features from the audio of the chunk (the input audio is processed using a plurality of convolutional layers to produce probabilities of redactions, paragraphs [0113-0115]).
In regard to claim 18, Huffman discloses the instructions cause one or more processors to perform or control performance of a further operation comprising performing pre-processing that includes one or more of:
removing background noise from the audio stream;
converting the audio stream into a particular digital format;
extracting Mel-Frequency Cepstral Coefficients (MFCCs) from the audio stream; and
combinations thereof (the audio stream is converted to a specific format, including MFCC representations of the audio stream, paragraph [0087]).
In regard to claim 20, Huffman discloses non-transitory computer-readable medium with instructions stored thereon that, when executed by a processor, cause the processor to perform or control performance of operations (paragraph [0224]) comprising:
obtaining a real-time audio stream of voice chat communication (audio within a real-time voice chat is obtained, paragraph [0087]);
dividing the real-time audio stream into a plurality of chunks, wherein each chunk comprises audio from the real-time audio stream from a respective time window (the input audio utilizes an input buffer to predict whether the buffer contains portions of content to be redacted, paragraph [0088]);
for each chunk,
providing the audio of the chunk to a machine-learning based detection model (the buffered audio is provided to a prediction model that predicts whether to-be-redacted content was spoken, paragraphs [0089-0091]);
obtaining, as output of the detection model, an indication whether policy-violating content is present in the audio of the chunk, wherein the detection model comprises an encoder that encodes the audio of the chunk into a feature vector and a classifier that generates the indication of whether policy-violating content is present in the audio of the chunk based on the feature vector (the first stage of the prediction model encodes the audio into an ordered sequence of distributions over phoneme probabilities parameterized by model weight vectors, paragraph [0089]; the sequence of phoneme distributions is compared to a lexicon of to-be-redacted words to determine whether to-be-redacted words were spoken or soon-to-be spoken in the audio of the input buffer, paragraphs [0090-0091]), and wherein, when the indication is that policy-violating content is present in the audio of the chunk, the detection model further outputs a time-span within the chunk that corresponds to the policy-violating content, and wherein the portion of the audio of the chunk is from within the time-span (see Fig. 1B, a time period t2-t3 is determined to include a target word to be redacted, paragraph [0112]);
in response to the indication that policy-violating content is present in the audio of the chunk, modifying the audio of the chunk to at least one of: mute at least a portion of the audio of the chunk or replace the at least a portion of the audio of the chunk with different audio, wherein the policy-violating content is a word (when the audio content is determined to likely include to-be-redacted words, the audio samples are set to zero before being copied to an output buffer, paragraph [0091]) ; and
providing the audio stream having the chunk with the modified audio to a client device for playback at the client device, wherein the audio stream is provided with a delay (the redacted audio stream is provided to an operator of a computer, Fig. 1A, 121, 122, 123, paragraph [0100]; with an introduced latency, paragraphs [0102-0103]), wherein the delay is greater than or equal a sum of the portion within the time-span and an inference time of the detection model for the chunk (a determination is made whether the full duration of a to-be-redacted target word has been spoken and the introduced latency is greater than the audio interval, paragraphs [0091] and [0103]; this will inherently be greater than or equal to the duration of a portion of the whole word and the introduced latency).
Huffman does not disclose dividing the real-time audio stream into the plurality of chunks comprises dynamically adjusting chunk sizes based on speech dynamics and environmental noise levels.
Wang discloses a method for dividing a real-time audio stream into a plurality of chunks that comprises dynamically adjusting chunk sizes based on speech dynamics and environmental noise levels (speech segments are determined by performing audio activity detection and determining a crossing rate indicative of speech, paragraph [0027]; the audio activity detection using environmental noise levels to first identify frames of noisy audio, and then merging frames into variable sized segments, paragraphs [0029-0031]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to divide the real-time audio stream into dynamically adjusted chunk sizes based on speech dynamics and environmental noise levels, because it would allow the audio stream to be segmented in practical noisy environments with low SNR, as taught by Wang (paragraph [0044]).
Huffman and Wang do not disclose the portion of the audio of the chunk is a portion within the time-span that corresponds to a last syllable of the word, and wherein replacing the portion of the audio of the chunk with different audio includes overlaying the detected policy-violating content with alternative audio.
Pearce discloses a method for replacing policy-violating content in an audio signal, wherein a portion of an audio chunk is a portion within the time-span that corresponds to a last syllable of the word, and wherein replacing the portion of the audio of the chunk with different audio includes overlaying the detected policy-violating content with alternative audio (a sound replacement algorithm replaces the last syllable of a policy-violating word with alternative audio, paragraph [0033]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to replace the last syllable of the word by overlaying the detected policy-violating content with alternative audio, because modifying only the last syllable of the word would allow the original audio to be modified with a reduced delay, as taught by Pearce (paragraph [0034]).
Claim(s) 3 and 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Huffman, in view of Wang, in further view of Pierce, and further in view of Audhkhasi et al. (U.S. Patent Application Pub. No. 2022/0310074, hereinafter “Audhkhasi”).
In regard to claims 3 and 17, although Huffman suggests transformer-type models may be used for prediction (which are based on self-attention), Huffman, Wang and Pierce do not expressly disclose the encoder further comprises one or more self-attention layers, wherein context from one or more prior chunks is provided to the self-attention layers.
Audhkhasi discloses a speech processing model comprising an encoder, wherein the encoder further comprises one or more self-attention layers, wherein context from one or more prior chunks is provided to the self-attention layers (Fig. 2, an audio encoder 300 includes a MiMo attention mechanism to compute an output sequence of feature vectors from an input sequence of feature vectors representing an audio signal over a plurality of T timesteps for generating a predicted feature vector at the current time step, paragraph [0042]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include one or more self-attention layers providing context from one or more prior chunks, because including context from one or more prior chunks via self-attention improves the prediction performance of the model, as taught by Audhkhasi (Fig. 4 and paragraph [0046]).
Claim(s) 5 and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Huffman, in view of Wang, in further view of Pierce, and further in view of Dantrey et al. (U.S. Patent No. 12,412,590, hereinafter “Dantrey”).
In regard to claims 5 and 19, Huffman, Wang and Pierce do not disclose the pre-processing is performed by a pre-processing layer of the detection model.
Dantrey et al. disclose a pre-processing layer of a neural network for removing noise from an audio signal (Fig. 3A, noise subtractor network 300 reduces noise energy in an input audio spectrogram, column 5, line 64 to column 6, line 10).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to add a pre-processing layer to the detection model, because such pre-processing layers generate enhanced audio by removing non-stationary background sounds in an online gaming environment, as taught by Dantrey (column 5, lines 34-63).
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 BRIAN LOUIS ALBERTALLI whose telephone number is (571)272-7616. The examiner can normally be reached M-F 8AM-3PM, 4PM-5PM.
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, Bhavesh Mehta can be reached at 571-272-7453. 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.
BLA 5/12/26
/BRIAN L ALBERTALLI/Primary Examiner, Art Unit 2656
Prosecution Insights