DETAILED ACTION
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 .
Claims 1-8 are presented for examination.
Response to Amendment
Applicant’s amendment has obviated the outstanding drawing, specification, and claim objections. Therefore, those objections are withdrawn (but see new grounds of objection infra).
Priority
Receipt is acknowledged of certified copies of papers required by 37 CFR 1.55.
Drawings
The drawings are objected to because in Figure 7, “head cicumcter” should be “head circumference” and “magmitiude response” in the top graph should be “magnitude response”, and the text that precedes both instances of “Hz” is illegible. Corrected drawing sheets in compliance with 37 CFR 1.121(d) are required in reply to the Office action to avoid abandonment of the application. Any amended replacement drawing sheet should include all of the figures appearing on the immediate prior version of the sheet, even if only one figure is being amended. The figure or figure number of an amended drawing should not be labeled as “amended.” If a drawing figure is to be canceled, the appropriate figure must be removed from the replacement sheet, and where necessary, the remaining figures must be renumbered and appropriate changes made to the brief description of the several views of the drawings for consistency. Additional replacement sheets may be necessary to show the renumbering of the remaining figures. Each drawing sheet submitted after the filing date of an application must be labeled in the top margin as either “Replacement Sheet” or “New Sheet” pursuant to 37 CFR 1.121(d). If the changes are not accepted by the examiner, the applicant will be notified and informed of any required corrective action in the next Office action. The objection to the drawings will not be held in abeyance.
Specification
The lengthy specification has not been checked to the extent necessary to determine the presence of all possible minor errors. Applicant’s cooperation is requested in correcting any errors of which applicant may become aware in the specification.
Claim Objections
Claim 1 is objected to because of the following informalities: “by processor” should be “by a processor” in the first recitation and “by the processor” in the second. Claims 2-8 are objected to for dependency on claim 1. Appropriate correction is required.
Claim Rejections - 35 USC § 102
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claims 1-2 and 4 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Bharitkar (WO 2020167309) (“Bharitkar”).
Regarding claim 1, Bharitkar discloses “[a] method for generating a personalized head-related transfer function (HRTF), the method comprising:
training, by [a] processor, a neural network model using multi-angle HRTFs labeled with body structure information of a learning object by means of a processor (compressed values at output of deepest encoder layer are used for training an artificial neural network to perform a function approximation task where the input to the artificial neural network is an angle [body information of the learning object] and the training data at the output of the artificial neural network is a latent representation of the deepest layer encoder output for the corresponding HRTF – Bharitkar, paragraph 27; neural network reconstructs HRTFs for arbitrary angles, for example, every 1 degree [i.e., the HRTF is constructed at multiple angles] – id. at paragraph 12; see also Fig. 6 (showing processor 602); see also paragraph 9 (disclosing that the HRTF is an impulse response corresponding to a lateral angle and elevation angle between the noise source and each ear [so each angle captures the structure of the environment including the position of the listener’s body])); and
obtaining, by [the] processor, multi-angle HRTFs at a time by inputting body structure information of a target user into the trained neural network model by means of the processor (direction parameter that may be azimuth and elevation angles [body information of a target user] describing a directionality of sound included in an audio signal are received, and the direction parameter is provided as input to the trained neural network to generate a compressed representation of a set of HRTFs [multi-angle HRTFs] – Bharitkar, paragraphs 45-47 and Fig. 5; see also paragraph 9 (disclosing that the HRTF is an impulse response corresponding to a lateral angle and elevation angle between the noise source and each ear [so each angle captures the structure of the environment including the position of the listener’s body])).”
Regarding claim 2, Bharitkar discloses that “the training of a neural network model includes applying supervised learning to the neural network model by setting the body structure information of the learning object as input data of the neural network model and setting the multi-angle HRTFs as output data of the neural network model (compressed values at output of deepest encoder layer are used for training an artificial neural network to perform a function approximation task where the input to the artificial neural network is an angle [body information of the learning object] and the training data at the output of the artificial neural network is a latent representation of the deepest layer encoder output for the corresponding HRTF – Bharitkar, paragraph 27; neural network reconstructs HRTFs for arbitrary angles, for example, every 1 degree [i.e., the HRTF is constructed at multiple angles] – id. at paragraph 12; see also paragraph 9 (disclosing that the HRTF is an impulse response corresponding to a lateral angle and elevation angle between the noise source and each ear [so each angle captures the structure of the environment including the position of the listener’s body])).”
Regarding claim 4, Bharitkar discloses that “the training of a neural network model includes training the neural network model such that the neural network model outputs the multi-angle HRTFs with reference to HRTFs of adjacent angles (compressed values at output of deepest encoder layer are used for training an artificial neural network to perform a function approximation task where the input to the artificial neural network is an angle [body information of the learning object] and the training data at the output of the artificial neural network is a latent representation of the deepest layer encoder output for the corresponding HRTF – Bharitkar, paragraph 27; neural network reconstructs HRTFs for arbitrary angles, for example, every 1 degree [i.e., adjacent angles] – id. at paragraph 12).”
Claim Rejections - 35 USC § 103
Claim 3 is rejected under 35 U.S.C. 103 as being unpatentable over Bharitkar in view of Li et al. (US 20200184339) (“Li”).
Regarding claim 3, the rejection of claim 1 is incorporated. Bharitkar further discloses that “the neural network model includes:
at least one fully connected layer [utilizing] … the body structure information of the learning object and the body structure information of the target user (artificial neural network may be a fully-connected neural network – Bharitkar, paragraph 28; compressed values at output of deepest encoder layer are used for training an artificial neural network to perform a function approximation task where the input to the artificial neural network is an angle [body structure informationof the learning object and user] and the training data at the output of the artificial neural network is a latent representation of the deepest layer encoder output for the corresponding HRTF –id. at paragraph 27; see also paragraph 9 (disclosing that the HRTF is an impulse response corresponding to a lateral angle and elevation angle between the noise source and each ear [so each angle captures the structure of the environment including the position of the listener’s body])); and
a … layer … outputting preset angle-specific HRTFs (compressed values at output of deepest encoder layer are used for training an artificial neural network to perform a function approximation task where the input to the artificial neural network is an angle and the training data at the output of the artificial neural network is a latent representation of the deepest layer encoder output for the corresponding HRTF [i.e., the layers of the network output HRTFs] –Bharitkar, paragraph 27; decoder reconstructs the HRTFs for arbitrary angles, for example, every 1 degree [so the HRTF outputs are preset to be specific to each possible integer angle] – id. at paragraph 12).”
Bharitkar appears not to disclose explicitly the further limitations of the claim. However, Li discloses that “the neural network model includes:
at least one … layer extracting features from the … information (topic embedding and word embedding are concatenated and fed into a CNN [comprising layers] for word and topic feature extraction – Li, paragraph 49); and
a bidirectional Long Short Term Memory (LSTM) layer receiving the extracted features as multiple inputs and outputting [data] (a bidirectional LSTM is employed for sequential [multiple] processing [including receipt] of the extracted features to obtain [output] a sentence representation [data] – Li, paragraph 49).”
Li and the instant application both relate to bidirectional LSTMs and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Bharitkar to employ a feature extraction layer that is passed to a bidirectional LSTM, as disclosed by Li, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow the data to be sequentially processed. See Li, paragraph 49.
Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Bharitkar in view of Refaat et al. (US 20220036186) (“Refaat”).
Regarding claim 5, the rejection of claim 1 is incorporated. Bharitkar further discloses that “the neural network model is trained such that a loss function … is minimized, … wherein
y
m
is a measured HRTF for an m-th angle and
y
^
m
is a predicted HRTF for the m-th angle (Bharitkar paragraphs 23-24 disclose the optimization of a cost [loss] function that includes a term corresponding to the sum of squared differences between the encoder input and the decoder output; paragraph 12 discloses that the decoder reconstructs HRTFs for arbitrary angles [i.e., the output of the decoder is the predicted HRTF for the mth angle, meaning that the input of the encoder whose output is the basis of the reconstruction is an original HRTF for the mth angle]).”
Bharitkar appears not to disclose explicitly the further limitations of the claim. However, Refaat discloses that “the neural network model is trained such that a loss function defined as the following [Equation 1] is minimized,
[Equation 1]
L
1
y
,
y
^
=
1
M
∑
m
=
0
M
-
1
{
y
m
-
y
^
m
}
2
(system trains the observation prediction neural network to minimize a mean squared error loss function between predicted observations and corresponding observations – Refaat, paragraph 71 [note that the claimed equation is the mean squared error]) ….”
Refaat and the instant application both relate to loss functions for neural network training and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Bharitkar to minimize a mean squared error loss function, as disclosed by Refaat, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would increase the accuracy of the predictions by allowing the network to learn from incorrect guesses. See Refaat, paragraph 71.
Claims 6-7 are rejected under 35 U.S.C. 103 as being unpatentable over Bharitkar in view of McDonald, Mean Absolute Log Error (MALE): A Better “Relative” Performance Metric, Medium, Jan. 18, 2023, https://medium.com/data-science/mean-absolute-log-error-male-a-better-relative-performance-metric-a8fd17bc5f75 (“McDonald”).
Regarding claim 6, the rejection of claim 1 is incorporated. Bharitkar further discloses that “the neural network model is trained such that a loss function … is minimized, … wherein
Y
m
is a measured HRTF at a frequency domain for an m-th angle and
Y
^
m
is a predicted HRTF at the frequency domain for the m-th angle (Bharitkar paragraphs 23-24 disclose the optimization of a cost [loss] function that includes a term corresponding to the sum of squared differences between the encoder input and the decoder output; paragraph 12 discloses that the decoder reconstructs HRTFs for arbitrary angles [i.e., the output of the decoder is the predicted HRTF for the mth angle, meaning that the input of the encoder whose output is the basis of the reconstruction is an original HRTF for the mth angle]; paragraph 34 discloses that the HRTF is computed in the frequency domain).”
Bharitkar appears not to disclose explicitly the further limitations of the claim. However, McDonald discloses that “a loss function defined as the following [Equation 2] is minimized,
[Equation 2]
L
2
Y
,
Y
^
=
1
M
∑
m
=
0
M
-
1
|
l
o
g
10
Y
m
-
l
o
g
10
Y
^
m
|
…. (McDonald, section entitled “… lead to better metrics” discloses the mean absolute log error, defined as
1
N
∑
t
=
1
T
|
l
o
g
(
f
t
y
t
)
|
, where f and y are forecast and observed values, respectively [corresponding to
Y
m
and
Y
^
m
, respectively; note that this is mathematically equivalent to the claimed equation because of the identity log (x/y) = log x – log y]).”
McDonald and the instant application both relate to error functions and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Bharitkar to use a mean absolute log error function as the loss function, as disclosed by McDonald, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would make the loss measure fairer by punishing underestimates and overestimates equally harshly. See McDonald, second bullet point under “Better error measures …”.
Regarding claim 7, the rejection of claim 1 is incorporated. Bharitkar further discloses that “the neural network model is trained such that a … loss function[] … is minimized, … wherein …
Y
m
is a measured HRTF at a frequency domain for an m-th angle and
Y
^
m
is a predicted HRTF at the frequency domain for the m-th angle (Bharitkar paragraphs 23-24 disclose the optimization of a cost [loss] function that includes a term corresponding to the sum of squared differences between the encoder input and the decoder output; paragraph 12 discloses that the decoder reconstructs HRTFs for arbitrary angles [i.e., the output of the decoder is the predicted HRTF for the mth angle, meaning that the input of the encoder whose output is the basis of the reconstruction is an original HRTF for the mth angle]; paragraph 34 discloses that the HRTF is computed in the frequency domain).”
Bharitkar appears not to disclose explicitly the further limitations of the claim. However, McDonald discloses that “a linear combination of first and second loss functions defined as the following [Equation 1] and [Equation 2], respectively, is minimized,
[Equation 1]
L
1
y
,
y
^
=
1
M
∑
m
=
0
M
-
1
{
y
m
-
y
^
m
}
2
[Equation 2]
L
2
Y
,
Y
^
=
1
M
∑
m
=
0
M
-
1
|
l
o
g
10
Y
m
-
l
o
g
10
Y
^
m
|
(McDonald, section entitled “… lead to better metrics” discloses the mean absolute log error, defined as
1
N
∑
t
=
1
T
|
l
o
g
(
f
t
y
t
)
|
, where f and y are forecast and observed values, respectively [corresponding to
Y
m
and
Y
^
m
, respectively; note that this is mathematically equivalent to the claimed equation because of the identity log (x/y) = log x – log y; note also that this is a linear combination of the two claimed loss functions under the special case where the coefficient corresponding to the first claimed loss function is zero]) ….” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Bharitkar to use a mean absolute log error function as the loss function, as disclosed by McDonald, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would make the loss measure fairer by punishing underestimates and overestimates equally harshly. See McDonald, second bullet point under “Better error measures …”.
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over Bharitkar in view of Milne et al. (US 20200107148) (“Milne”).
Regarding claim 8, the rejection of claim 1 is incorporated. Bharitkar further discloses a “neural network model” and “multi-angle HRTFs”, as shown in the rejection of claim 1.
Bharitkar appears not to disclose explicitly the further limitations of the claim. However, Milne discloses that “the obtaining of … HRTFs at a time includes obtaining … HRTFs at a time by further inputting an ear image of the target user into the … model (a person or an assistant takes one or more 3D images of the person’s ears and/or takes a panoramic set of images of each ear; the photo or photos are processed onboard the device taking the photos by inputting the photographs to a module [model] that generates or otherwise returns HRTFs – Milne, paragraph 43).”
Milne and the instant application both relate to HRTF generation and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Bharitkar to generate the HRTFs by taking photographs of the subject’s ears and inputting them into the model, as disclosed by Milne, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would save on equipment costs that would otherwise be incurred using other types of data such as sound data. See Milne, paragraph 3.
Response to Arguments
Applicant's arguments filed September 18, 2026 (“Remarks”) have been fully considered but they are not persuasive.
Applicant argues that Bharitkar does not anticipate amended claim 1 because Bharitkar does not use “anatomical dimensions” of a body as the network input variable. Remarks at 8-9. However, “anatomical dimensions” of a body are not claimed. Rather, the amended claims recite the considerably broader language “body structure information”. Since neither the specification nor the claims define “body structure information,” the term may be construed to mean any information about the structure of the body, including the ears’ orientation in space, which is necessarily influenced by structural factors such as the subject’s height, head circumference, etc. For instance, a taller person’s head will be positioned higher than a shorter person’s head when they both stand upright, which affects the angle between the ears and the noise source. Thus, the sound angle of Bharitkar implicitly includes information about the subject’s body structure, and indeed is itself structural information because the position of a body in space relates to the body’s structure.
Conclusion
THIS ACTION IS MADE FINAL. 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 RYAN C VAUGHN whose telephone number is (571)272-4849. The examiner can normally be reached M-R 7:00a-5:00p ET.
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, Kamran Afshar, can be reached at 571-272-7796. 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.
/RYAN C VAUGHN/ Primary Examiner, Art Unit 2125
