Abstract: A method of generating audio output includes displaying a graphical user interface (GUI) at a user device. The GUI represents an area having multiple regions and multiple audio capture devices are located in the area. The method also includes receiving audio data from the multiple audio capture devices. The method further includes receiving an input indicating a selected region of the multiple regions. The method also includes generating, at the user device, audio output based on audio data from a subset of the multiple audio capture devices. Each audio capture device in the subset is located in the selected region.

Abstract: A system includes a feedback filter configured to receive an audio signal from a microphone and to process the audio signal based on a feedback path between a loudspeaker and the microphone. The system also includes an adaptive filter coupled to the feedback filter. The system further includes a signal combiner coupled to the microphone and to the adaptive filter.

Abstract: A method of selectively authorizing access includes obtaining, at an authentication device, first information corresponding to first synthetic biometric data. The method also includes obtaining, at the authentication device, first common synthetic data and second biometric data. The method further includes generating, at the authentication device, second common synthetic data based on the first information and the second biometric data. The method also includes selectively authorizing, by the authentication device, access based on a comparison of the first common synthetic data and the second common synthetic data.

Abstract: A crosstalk cancelation technique reduces feedback in a shared acoustic space by canceling out some or all parts of sound signals that would otherwise be produced by a loudspeaker to only be captured by a microphone that, recursively, would cause these sounds signals to be reproduced again on the loudspeaker as feedback. Crosstalk cancelation can be used in a multichannel acoustic system (MAS) comprising an arrangement of microphones, loudspeakers, and a processor to together enhance conversational speech between in a shared acoustic space. To achieve crosstalk cancelation, a processor analyzes the inputs of each microphone, compares it to the output of far loudspeaker(s) relative to each such microphone, and cancels out any portion of a sound signal received by the microphone that matches signals that were just produced by the far loudspeaker(s) and sending only the remaining sound signal (if any) to such far loudspeakers.

Abstract: Disclosed is a feature extraction and classification methodology wherein audio data is gathered in a target environment under varying conditions. From this collected data, corresponding features are extracted, labeled with appropriate filters (e.g., audio event descriptions), and used for training deep neural networks (DNNs) to extract underlying target audio events from unlabeled training data. Once trained, these DNNs are used to predict underlying events in noisy audio to extract therefrom features that enable the separation of the underlying audio events from the noisy components thereof.

Abstract: An accessory device having multiple speakers and/or microphones to perform a number of audio functions, for use with mobile devices, are provided. The audio transducers (e.g., microphones and/or speakers) may be housed in one or more extendable and/or rotationally adjustable arms. To compensate for the unwanted signal feedback between the speakers and microphones, acoustic echo cancellation may be implemented to determine the proper distance and relative location between the speakers and microphones. Acoustic echo cancellation removes the echo from voice communications to improve the quality of the sound. The removal of the unwanted signals captured by the microphones may be accomplished by characterizing the audio signal paths from the speakers to the microphones (speaker-to-microphone path distance profile), including the distance and relative location between the speakers and microphones.

Abstract: A method of performing noise reduction includes capturing a first audio signal at a first microphone of a first device. The method also includes receiving, at the first device, audio data representative of a second audio signal from a second device. The second audio signal is captured by a second microphone of the second device. The method further includes performing noise reduction on the first audio signal based at least in part on the audio data representative of the second audio signal.

Abstract: A method of generating audio output includes displaying a graphical user interface (GUI) at a user device. The GUI represents an area having multiple regions and multiple audio capture devices are located in the area. The method also includes receiving audio data from the multiple audio capture devices. The method further includes receiving an input indicating a selected region of the multiple regions. The method also includes generating, at the user device, audio output based on audio data from a subset of the multiple audio capture devices. Each audio capture device in the subset is located in the selected region.

Abstract: Techniques for processing directionally-encoded audio to account for spatial characteristics of a listener playback environment are disclosed. The directionally-encoded audio data includes spatial information indicative of one or more directions of sound sources in an audio scene. The audio data is modified based on input data identifying the spatial characteristics of the playback environment. The spatial characteristics may correspond to actual loudspeaker locations in the playback environment. The directionally-encoded audio may also be processed to permit focusing/defocusing on sound sources or particular directions in an audio scene. The disclosed techniques may allow a recorded audio scene to be more accurately reproduced at playback time, regardless of the output loudspeaker setup. Another advantage is that a user may dynamically configure audio data so that it better conforms to the user's particular loudspeaker layouts and/or the user's desired focus on particular subjects or areas in an audio scene.

Abstract: A method for audio signal processing is described. The method includes decomposing a recorded auditory scene into a first category of localizable sources and a second category of ambient sound. The method also includes recording an indication of the directions of each of the localizable sources. The method may be performed with a device having a microphone array.

Abstract: A method for speech restoration by an electronic device is described. The method includes obtaining a noisy speech signal. The method also includes suppressing noise in the noisy speech signal to produce a noise-suppressed speech signal. The noise-suppressed speech signal has a bandwidth that includes at least three subbands. The method further includes iteratively restoring each of the at least three subbands. Each of the at least three subbands is restored based on all previously restored subbands of the at least three subbands.

Abstract: One example device includes a camera; a display device; a memory; and a processor in communication with the memory to receive audio signals from two or more microphones or a far-end device; receive first location information and second location information, the first location information for a visual identification of an audio source of the received audio signals and the second location information identifying a direction of arrival from the audio source; receive a first adjustment to a first portion of a UI to change either a visual identification or a coordinate direction of a direction focus; in response to the first adjustment, automatically perform a second adjustment to a second portion of the UI to change the other of the visual identification or the coordinate direction of the direction focus; and process the audio signals to filter sounds outside the direction focus, or emphasize sounds within the direction focus.

Abstract: Methods, systems and articles of manufacture for recognizing and locating one or more objects in a scene are disclosed. An image and/or video of the scene are captured. Using audio recorded at the scene, an object search of the captured scene is narrowed down. For example, the direction of arrival (DOA) of a sound can be determined and used to limit the search area in a captured image/video. In another example, keypoint signatures may be selected based on types of sounds identified in the recorded audio. A keypoint signature corresponds to a particular object that the system is configured to recognize. Objects in the scene may then be recognized using a shift invariant feature transform (SIFT) analysis comparing keypoints identified in the captured scene to the selected keypoint signatures.

Abstract: A method for echo reduction by an electronic device is described. The method includes nulling at least one speaker. The method also includes mixing a set of runtime audio signals based on a set of acoustic paths to determine a reference signal. The method also includes receiving at least one composite audio signal that is based on the set of runtime audio signals. The method further includes reducing echo in the at least one composite audio signal based on the reference signal.

Abstract: A wireless device is described. The wireless device includes at least two microphones on the wireless device. The microphones are configured to capture sound from a target user. The wireless device also includes processing circuitry. The processing circuitry is coupled to the microphones. The processing circuitry is configured to locate the target user. The wireless device further includes a communication interface. The communication interface is coupled to the processing circuitry. The communication interface is configured to receive external device microphone audio from at least one external device microphone to assist the processing circuitry in the wireless device to locate the target user.

Abstract: Disclosed is a feature extraction and classification methodology wherein audio data is gathered in a target environment under varying conditions. From this collected data, corresponding features are extracted, labeled with appropriate filters (e.g., audio event descriptions), and used for training deep neural networks (DNNs) to extract underlying target audio events from unlabeled training data. Once trained, these DNNs are used to predict underlying events in noisy audio to extract therefrom features that enable the separation of the underlying audio events from the noisy components thereof.

Abstract: A method for speech restoration by an electronic device is described. The method includes obtaining a noisy speech signal. The method also includes suppressing noise in the noisy speech signal to produce a noise-suppressed speech signal. The noise-suppressed speech signal has a bandwidth that includes at least three subbands. The method further includes iteratively restoring each of the at least three subbands. Each of the at least three subbands is restored based on all previously restored subbands of the at least three subbands.

Abstract: A method of selectively authorizing access includes obtaining, at an authentication device, first information corresponding to first synthetic biometric data. The method also includes obtaining, at the authentication device, first common synthetic data and second biometric data. The method further includes generating, at the authentication device, second common synthetic data based on the first information and the second biometric data. The method also includes selectively authorizing, by the authentication device, access based on a comparison of the first common synthetic data and the second common synthetic data.

Abstract: A wireless device is provided that makes use of other nearby audio transducer devices to generate a surround sound effect for a targeted user. To do this, the wireless device first ascertains whether there are any nearby external microphones and/or loudspeaker devices. An internal microphone for the wireless device and any other nearby external microphones may be used to ascertain a location of the desired/targeted user as well as the nearby loudspeaker devices. This information is then used to generate a surround sound effect for the desired/targeted user by having the wireless device steer audio signals to its internal loudspeakers and/or the nearby external loudspeaker devices.

Abstract: Techniques for processing directionally-encoded audio to account for spatial characteristics of a listener playback environment are disclosed. The directionally-encoded audio data includes spatial information indicative of one or more directions of sound sources in an audio scene. The audio data is modified based on input data identifying the spatial characteristics of the playback environment. The spatial characteristics may correspond to actual loudspeaker locations in the playback environment. The directionally-encoded audio may also be processed to permit focusing/defocusing on sound sources or particular directions in an audio scene. The disclosed techniques may allow a recorded audio scene to be more accurately reproduced at playback time, regardless of the output loudspeaker setup. Another advantage is that a user may dynamically configure audio data so that it better conforms to the user's particular loudspeaker layouts and/or the user's desired focus on particular subjects or areas in an audio scene.