Marvell International Limited

Marvellworld Trade Ltd.

Embodyvr, Inc.

Marvell Asia Pte., Ltd.

Kapil Jain

Jin Xie

Zining Wu

Sriharsha Annadore

Kaushik Sunder

Erfan Soltanmohammadi

Shruti Badhwar

Nikhil Ratnesh Javeri

Faiyadh Shahid

Sungyub Daniel Yoo

Sasan Cyrusian

Song Chen

Wenzong Pan

Yongxu Wang

Xiaoxiao Zhao

Wenrong Qian

Sungyub D Yoo

Marielle Venita Jakobsons

Abhilash Mathew

Yuxiang Wang

Christopher Painter

Cai Wang

Yafei Hu

Haiying Xia

Liu Han

Kieran Coulter

A system comprises a sound source to output an audio signal to a plurality of speakers arranged in a plurality of locations in a vehicle. A laser device is disposed on an object representing a human head placed in a seat of the vehicle to scan the locations of the speakers. A microphone is disposed in each ear of the object. A controller routes the audio signal to the speakers one speaker at a time, receives audio signals received via the microphones, and compiles binaural acoustic data for the vehicle based on the audio signals received via the microphones. The controller receives scan data from the laser device and generates geometric data for the vehicle based on the scan data. The controller generates head-related transfer functions (HRTFs) for the object based on the binaural acoustic data and the geometric data for the vehicle.

System and method to virtually mix and audition audio content for vehicles

Methods, systems, and program products for generating a virtual studio are disclosed. In one embodiment a method includes processing image information for at least one pinna of a user to generate a head-related transfer function (HRTF) profile of the user. A studio model that includes a studio-specific acoustic profile is accessed such as by a virtual studio client application executing on a laptop. An audio configuration of the studio model is selected based on the studio-specific acoustic profile. An audio media source is activated and the audio configuration is applied in combination with the HRTF profile of the user to audio generated by the audio media source.

Personalized and integrated virtual studio

An application such as a machine learning workflow manager executes a machine learning workflow utilizing varying sets of parameters and tracks the utilized parameters and performance metrics for each execution. The manager generates a unique experiment identifier (ID) for each set of parameters used for executing a workflow and stores the experiment ID and the set of parameters along with results of the execution, e.g., performance metrics, output data, program code, etc. In some implementations, the manager can algorithmically generate sets of parameters for experiments of the workflow. Once experimentation for the workflow is complete, the manager utilizes the stored performance metrics to identify an experiment which exhibited the best performance and can retrieve the set of parameters, output data, or generated program code for deployment of the workflow using the associated experiment ID.

Automated versioning and evaluation of machine learning workflows

A first microphone on a first earpiece of a personal audio delivery device measures a first head related transfer function (HRTF) associated with a sound source. A second microphone on an earpiece of a personal audio delivery device measures a second HRTF associated with the sound source. An interaural time difference (ITD) is determined based on the first HRTF and second HRTF. A determination is made that the sound source is located in a first region based on the interaural time difference. A determination is made that the sound source is located in a second region within the first region based on the ITD of the first HRTF and the second HRTF and an ITD associated with third HRTFs for the second region. A location of the sound source is determined within the second region, which in some examples, is improved with a Kalman filter model.

Binaural sound source localization

A video is received from a video capture device. The video capture device has a front facing camera and a display screen which displays the video captured by the video capture device in real time to a user. One or more images of a pinna and head of the user in the video are used to automatically determine one or more features associated with the user. The one or more features include an anatomy of the user, a demographic of the user, a latent feature of the user, and an indication of an accessory worn by the user. Based on the one or more features and one or more HRTF models, a head related transfer function (HRTF) is determined which is personalized to the user.

Personalized head related transfer function (HRTF) based on video capture

An image of a pinna is captured. Based on the image of the pinna, a non-linear transfer function is determined which characterizes how sound is transformed at the pinna. A signal is output indicative of one or more audio cues to facilitate spatial localization of sound via the pinna, where the one or more audio cues is based on the non-linear transfer function.

System and method to capture image of pinna and characterize human auditory anatomy using image of pinna

Aspects of the disclosure include an apparatus that has a first clock generator and a second clock generator. The first clock generator is configured to drive a first circuit, causing the first circuit to (i) receive a signal corresponding to an audio input, and (ii) determine whether an energy level of the signal exceeds a predetermined threshold. The second clock generator is activated when the first circuit determines that the energy level of the signal exceeds the predetermined threshold. The second clock generator is configured to drive a second circuit, causing the second circuit to determine whether the signal matches a predetermined pattern. A third circuit is activated when the second circuit determines that the signal matches the predetermined pattern.

Apparatus and method for activating circuits

A signal indicative of sound detected by at least one sensor at an audio device is received. The audio device may at least partially covers a pinna and the detected sound may interact with at least a torso of a human body, but can also interact with the head and shoulder. The signal is modulated with a non-linear transfer function to generate a modulated signal indicative of one or more audio cues for spatializing the detected sound while the audio device at least partially covers a pinna. Sound is output by the audio device based on the modulated signal.

Spatially ambient aware personal audio delivery device

An image sensor in a first earcup captures an image of a pinna. First sound is output by a transducer in a second earcup located at the pinna and respective second sound is detected by each of one or more microphones in the second earcup located at the pinna. Based on the captured image and the respective second audio sound from each of the one or more microphones, a non-linear transfer function is determined which characterizes how sound is transformed by the pinna. A signal is generated indicative of one or more audio cues for spatializing third sound based on the determined non-linear transfer function.

Image and audio based characterization of a human auditory system for personalized audio reproduction

A first sound is output by a first transducer in an earcup located at a pinna. Respective second sound based on the output of the first audio sound is detecting by each of one or more microphones in the earcup located at the pinna. Based on the respective detected second sound, a non-linear transfer function is determined which characterizes how sound is transformed via the pinna. A signal indicative of one or more audio cues for spatializing third sound is generated based on the determined non-linear transfer function.

Audio based characterization of a human auditory system for personalized audio reproduction

Embodiments include a method comprising generating, based on receiving audio signals from a sound source and a noise source, (i) a first input signal and (ii) a second input signal; generating, based on the first input signal and the second input signal, an average signal; operating a first phase loop by phase shifting the first input signal to generate a first intermediate signal such that a sound component in the first intermediate signal is substantially phase aligned with a sound component in the average signal; operating a second phase loop by phase shifting the second input signal to generate a second intermediate signal such that a sound component in the second intermediate signal is substantially phase aligned with the sound component in the average signal; and generating, based on the first intermediate signal and the second intermediate signal, an output audio signal that comprises audio signals from the sound source.

Method and apparatus for tracking sound source movement for audio signal processing

A method includes determining a preprocessed audio signal by removing some noise from an input audio signal. Here, portions of the preprocessed audio signal that include speech are separated by portions of the preprocessed audio signal that include residual noise. Additionally, the method includes determining an amplified signal by suppressing the preprocessed audio signal over the portions that include residual noise, and maintaining the preprocessed audio signal over the portions that include speech.

Residual noise suppression

Systems, methods, and other embodiments associated with a processor that includes selectively enabled features are described. According to one embodiment, a processor includes a plurality of processing routines embedded within the processor that when executed cause the processor to implement corresponding processor features. The processor includes a processor engine configured to determine whether a processing routine of the plurality of processing routines is enabled based, at least in part, on a corresponding value in a control register. The processing engine is configured to selectively execute the processing routine based, at least in part, on whether the value indicates that the processing routine is enabled.

ROM-code programmable digital signal processor

A magnetic sensor mounted on a headband of the personal audio delivery device may output a sensor signal indicative of an interaction between a magnetic field of a transducer of a personal audio delivery device and the magnetic sensor. A head size of a head on which the personal audio delivery device is worn is calculated based on the sensor signal from the magnetic sensor. Based on the head size, a non-linear transfer function is identified which characterizes how sound is transformed via the head with the calculated head size. An output signal is generated indicative of one or more audio cues to facilitate spatialization of sound associated with the output signal based on the identified non-linear transfer function. The sound associated with the output signal is output by the transducer of the personal audio delivery device.

Growing community of inventors

Santa Clara, CA, United States of America

Kapil Jain