Patents by Inventor Milind Borkar
Milind Borkar has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Patent number: 11783841Abstract: A method and system for secure speaker authentication between a caller device and a first device using an authentication server are provided. The system comprises extracting features into a feature matrix from an incoming audio call; generating a partial i-vector, wherein the partial i-vector includes a first low-order statistic; sending the partial i-vector to the authentication server; and receiving from the authentication server a match score generated based on a full i-vector and another i-vector being stored on the authentication server, wherein the full i-vector is generated from the partial i-vector.Type: GrantFiled: March 15, 2021Date of Patent: October 10, 2023Assignee: ILLUMA LABS INC.Inventor: Milind Borkar
-
Patent number: 11699445Abstract: A system and method for improving T-matrix training for speaker recognition, comprising receiving an audio input, divisible into a plurality of audio frames including at least an audio sample of a human speaker; generating for each audio frame a feature vector; generating for a first plurality of feature vectors centered statistics of at least a zero order and a first order; generating a first i-vector, the first i-vector representing the human speaker; and generating an optimized T-matrix training sequence computation, based on at least the first i-vector.Type: GrantFiled: March 15, 2021Date of Patent: July 11, 2023Assignee: ILLUMA LABS INC.Inventor: Milind Borkar
-
Patent number: 11521622Abstract: A system and method for efficient universal background model (UBM) training for speaker recognition, including: receiving an audio input, divisible into a plurality of audio frames, wherein at least a first audio frame of the plurality of audio frames includes an audio sample having a length above a first threshold extracting at least one identifying feature from the first audio frame and generating a feature vector based on the at least one identifying feature; generating an optimized training sequence computation based on the feature vector and a Gaussian Mixture Model (GMM), wherein the GMM is associated with a plurality of components, wherein each of the plurality of components is defined by a covariance matrix, a mean vector, and a weight vector; and updating any of the associated components of the GMM based on the generated optimized training sequence computation.Type: GrantFiled: October 27, 2020Date of Patent: December 6, 2022Assignee: ILLUMA Labs Inc.Inventor: Milind Borkar
-
Publication number: 20210201919Abstract: A method and system for secure speaker authentication between a caller device and a first device using an authentication server are provided. The system comprises extracting features into a feature matrix from an incoming audio call; generating a partial i-vector, wherein the partial i-vector includes a first low-order statistic; sending the partial i-vector to the authentication server; and receiving from the authentication server a match score generated based on a full i-vector and another i-vector being stored on the authentication server, wherein the full i-vector is generated from the partial i-vector.Type: ApplicationFiled: March 15, 2021Publication date: July 1, 2021Applicant: ILLUMA Labs Inc.Inventor: Milind BORKAR
-
Publication number: 20210201917Abstract: A system and method for improving T-matrix training for speaker recognition, comprising receiving an audio input, divisible into a plurality of audio frames including at least an audio sample of a human speaker; generating for each audio frame a feature vector; generating for a first plurality of feature vectors centered statistics of at least a zero order and a first order; generating a first i-vector, the first i-vector representing the human speaker; and generating an optimized T-matrix training sequence computation, based on at least the first i-vector.Type: ApplicationFiled: March 15, 2021Publication date: July 1, 2021Applicant: ILLUMA Labs Inc.Inventor: Milind BORKAR
-
Patent number: 10950244Abstract: A system and method for enrolling a speaker in a speaker authentication and identification system (AIS), the method comprising: generating a user account, the user account comprising: a user identifier based on one or more metadata elements associated with an audio input received from an end device; generating a first i-vector from an audio frame of the audio input, a trained T-matrix, and a Universal Background Model (UBM), wherein the first i-vector generation comprises an optimized computation; and associating the user account with the first i-vector.Type: GrantFiled: April 16, 2019Date of Patent: March 16, 2021Assignee: ILLUMA Labs LLC.Inventor: Milind Borkar
-
Patent number: 10950243Abstract: A system and method for improving T-matrix training for speaker recognition are provided. The method includes receiving an audio input, divisible into a plurality of audio frames, wherein at least a first audio frame includes an audio sample of a human speaker, the sample having a length above a first threshold; generating for each audio frame a feature vector; generating for a first plurality of feature vectors centered statistics of at least a zero order and a first order; generating a first i-vector, the first i-vector representing the human speaker; generating an optimized T-matrix training sequence computation, based on the first i-vector, an initialized T-matrix, the centered statistics, and a Gaussian mixture model (GMM) of a trained universal background model (UBM).Type: GrantFiled: March 1, 2019Date of Patent: March 16, 2021Assignee: ILLUMA Labs Inc.Inventor: Milind Borkar
-
Publication number: 20210043215Abstract: A system and method for efficient universal background model (UBM) training for speaker recognition, including: receiving an audio input, divisible into a plurality of audio frames, wherein at least a first audio frame of the plurality of audio frames includes an audio sample having a length above a first threshold extracting at least one identifying feature from the first audio frame and generating a feature vector based on the at least one identifying feature; generating an optimized training sequence computation based on the feature vector and a Gaussian Mixture Model (GMM), wherein the GMM is associated with a plurality of components, wherein each of the plurality of components is defined by a covariance matrix, a mean vector, and a weight vector; and updating any of the associated components of the GMM based on the generated optimized training sequence computation.Type: ApplicationFiled: October 27, 2020Publication date: February 11, 2021Applicant: ILLUMA Labs Inc.Inventor: Milind BORKAR
-
Patent number: 10832683Abstract: A system and method for efficient universal background model (UBM) training for speaker recognition, including: receiving an audio input, divisible into a plurality of audio frames, wherein at least a first audio frame of the plurality of audio frames includes an audio sample having a length above a first threshold extracting at least one identifying feature from the first audio frame and generating a feature vector based on the at least one identifying feature; generating an optimized training sequence computation based on the feature vector and a Gaussian Mixture Model (GMM), wherein the GMM is associated with a plurality of components, wherein each of the plurality of components is defined by a covariance matrix, a mean vector, and a weight vector; and updating any of the associated components of the GMM based on the generated optimized training sequence computation.Type: GrantFiled: November 28, 2018Date of Patent: November 10, 2020Assignee: ILLUMA Labs LLC.Inventor: Milind Borkar
-
Publication number: 20190244622Abstract: A system and method for enrolling a speaker in a speaker authentication and identification system (AIS), the method comprising: generating a user account, the user account comprising: a user identifier based on one or more metadata elements associated with an audio input received from an end device; generating a first i-vector from an audio frame of the audio input, a trained T-matrix, and a Universal Background Model (UBM), wherein the first i-vector generation comprises an optimized computation; and associating the user account with the first i-vector.Type: ApplicationFiled: April 16, 2019Publication date: August 8, 2019Applicant: ILLUMA Labs Inc.Inventor: Milind BORKAR
-
Publication number: 20190198025Abstract: A system and method for improving T-matrix training for speaker recognition are provided. The method includes receiving an audio input, divisible into a plurality of audio frames, wherein at least a first audio frame includes an audio sample of a human speaker, the sample having a length above a first threshold; generating for each audio frame a feature vector; generating for a first plurality of feature vectors centered statistics of at least a zero order and a first order; generating a first i-vector, the first i-vector representing the human speaker; generating an optimized T-matrix training sequence computation, based on the first i-vector, an initialized T-matrix, the centered statistics, and a Gaussian mixture model (GMM) of a trained universal background model (UBM).Type: ApplicationFiled: March 1, 2019Publication date: June 27, 2019Applicant: ILLUMA Labs Inc.Inventor: Milind BORKAR
-
Publication number: 20190164557Abstract: A system and method for efficient universal background model (UBM) training for speaker recognition, including: receiving an audio input, divisible into a plurality of audio frames, wherein at least a first audio frame of the plurality of audio frames includes an audio sample having a length above a first threshold extracting at least one identifying feature from the first audio frame and generating a feature vector based on the at least one identifying feature; generating an optimized training sequence computation based on the feature vector and a Gaussian Mixture Model (GMM), wherein the GMM is associated with a plurality of components, wherein each of the plurality of components is defined by a covariance matrix, a mean vector, and a weight vector; and updating any of the associated components of the GMM based on the generated optimized training sequence computation.Type: ApplicationFiled: November 28, 2018Publication date: May 30, 2019Applicant: ILLUMA Labs Inc.Inventor: Milind BORKAR
-
Patent number: 9837971Abstract: For protecting a speaker, an input signal is received, and an excursion of the speaker that would be caused by the input signal is predicted. In response to the predicted excursion exceeding a threshold, a targeted excursion of the speaker is determined by compressing the predicted excursion. The targeted excursion is translated into an output signal, which is output to the speaker.Type: GrantFiled: April 16, 2012Date of Patent: December 5, 2017Assignee: TEXAS INSTRUMENTS INCORPORATEDInventors: Chenchi Luo, Milind A. Borkar, Arthur J. Redfern
-
Patent number: 9007334Abstract: An embodiment of the invention provides a method of creating a statistical model of a baseline capacitance CP of a capacitive sensor located on a capacitive-touch screen. A sensed capacitance CS of a capacitive sensor is measured during a particular state of the electronic device that includes the capacitive-touch screen. When physical contact is not made with the capacitive sensor, the sensed capacitance CS is stored as a baseline capacitance CP. The baseline capacitance CP is then used to create the statistical model for that particular state of the electronic device. When physical contact is made with the capacitive sensor, the value of the baseline capacitance CP of the capacitive sensor is subtracted from the value of the sensed capacitance CS and the result, CF=(CS?CP), is sent to a touch detection circuit.Type: GrantFiled: June 7, 2012Date of Patent: April 14, 2015Assignee: Texas Instruments IncorporatedInventors: Chenchi Eric Luo, Milind Borkar
-
Publication number: 20140028605Abstract: An embodiment of the invention provides a method and apparatus for determining what type of interaction is made with a capacitive touch screen. A capacitance sensor with the largest sensed capacitance in a group of capacitance sensors is determined. Next, a parametric surface is determined from the value of the largest sensed capacitance and the values of the sensed capacitances in the group of capacitance sensors. From the parametric surface, an interpolated peak capacitance, a curvature K at the interpolated peak and an orientation ? at the interpolated peak are determined. Based on the interpolated peak capacitance, the curvature K and the orientation ?, the type of interaction made with the capacitive-touch screen is identified.Type: ApplicationFiled: July 26, 2012Publication date: January 30, 2014Applicant: TEXAS INSTRUMENTS INCORPORATEDInventors: Chenchi Eric Luo, Milind Borkar
-
Publication number: 20130328822Abstract: An embodiment of the invention provides a method of creating a statistical model of a baseline capacitance CP of a capacitive sensor located on a capacitive-touch screen. A sensed capacitance CS of a capacitive sensor is measured during a particular state of the electronic device that includes the capacitive-touch screen. When physical contact is not made with the capacitive sensor, the sensed capacitance CS is stored as a baseline capacitance CP. The baseline capacitance CP is then used to create the statistical model for that particular state of the electronic device. When physical contact is made with the capacitive sensor, the value of the baseline capacitance CP of the capacitive sensor is subtracted from the value of the sensed capacitance CS and the result, CF=(CS?CP), is sent to a touch detection circuit.Type: ApplicationFiled: June 7, 2012Publication date: December 12, 2013Applicant: TEXAS INSTRUMENTS INCORPORATEDInventors: Chenchi Eric Luo, Milind Borkar
-
Publication number: 20130022207Abstract: For protecting a speaker, an input signal is received and filtered into component signals. A sum of the component signals is approximately equal to the input signal. The component signals include at least first and second component signals. A perceived loudness to a human from the speaker is more sensitive to the first component signal than to the second component signal. A temperature of the speaker is estimated. In response to the estimated temperature, the second component signal is scaled. An output signal is output to the speaker in response to the first component signal and the scaled second component signal.Type: ApplicationFiled: April 16, 2012Publication date: January 24, 2013Applicant: Texas Instruments IncorporatedInventors: Chenchi Luo, Milind A. Borkar, Arthur J. Redfern
-
Publication number: 20120281844Abstract: For protecting a speaker, an input signal is received, and an excursion of the speaker that would be caused by the input signal is predicted. In response to the predicted excursion exceeding a threshold, a targeted excursion of the speaker is determined by compressing the predicted excursion. The targeted excursion is translated into an output signal, which is output to the speaker.Type: ApplicationFiled: April 16, 2012Publication date: November 8, 2012Applicant: Texas Instruments IncorporatedInventors: Chenchi LUO, Milind A. BORKAR, Arthur J. REDFERN
-
Patent number: 8306149Abstract: An apparatus is provided. In the apparatus, an input to index (I2I) module maps a complex input into a real signal. A real data tap delay line is coupled to the I2I module and includes N delay-elements. A complex data tap delay line is configured to receive the complex input and includes M delay elements. A set of K of non-linear function modules is also provided. Each non-linear function module from the set has at least one real input, at least one complex input, and at least one complex output. A configurable connectivity crossbar multiplexer couples K of the N real tap delay line elements to real inputs of the set non-linear functions and couples K of the M complex tap delay line elements to complex inputs of the set non-linear function modules.Type: GrantFiled: October 1, 2009Date of Patent: November 6, 2012Assignee: Texas Instruments IncorporatedInventors: Fernando Alberto Mujica, Hardik Prakash Gandhi, Lei Ding, Milind Borkar, Zigang Yang, Roland Sperlich, Lars Morten Jorgensen, William L. Abbott
-
Patent number: 8306150Abstract: Systems and methods for identifying a transmission channel response and a feedback channel response from a plurality of composite system responses are disclosed. A plurality of shifted feedback signals are created by shifting a feedback signal frequency by a plurality of first offset values and/or by shifting a transmission signal frequency by a plurality of second offset values. The feedback signals are compared to an input signal to identify the transmission channel response and/or a feedback channel response. A control signal is generated for a pre-distortion circuit to modify the input signal by an inverse of the transmission channel response. The composite system response is measured at a plurality of operating frequencies and at the plurality of offset values. The measurements are stored in a matrix and singular value decomposition is applied to the matrix of measurements to calculate the transmission channel response and feedback channel response.Type: GrantFiled: June 25, 2010Date of Patent: November 6, 2012Assignee: Texas Instruments IncorporatedInventors: Fernando A. Mujica, Carson A. Wick, Lei Ding, Milind Borkar, Roland Sperlich