Techniques are disclosed for scene analysis including the use of acoustic imaging and computer audio vision processes for monitoring applications. In some embodiments, an acoustic image device is utilized with a microphone array, image sensor, acoustic image controller, and a controller. In some cases, the controller analyzes at least a portion of the spatial spectrum within the acoustic image data to detect sound variations by identifying regions of pixels having intensities exceeding a particular threshold. In addition, the controller can detect two or more co-occurring sound events based on the relative distance between pixels with intensities exceeding the threshold. The resulting data fusion of image pixel data, audio sample data, and acoustic image data can be analyzed using computer audio vision, sound/voice recognition, and acoustic signature techniques to recognize/identify audio and visual features associated with the event and to empirically or theoretically determine one or more conditions causing each event.

Systems and methods for sound source detection and localization utilizing an autonomous driving vehicle (ADV) are disclosed. The method includes receiving audio data from a number of audio sensors mounted on the ADV. The audio data comprises sounds captured by the audio sensors and emitted by one or more sound sources. Based on the received audio data, the method further includes determining a number of sound source information. Each sound source information comprises a confidence score associated with an existence of a specific sound. The method further includes generating a data representation to report whether there exists the specific sound within the driving environment of the ADV. The data representation comprises the determined sound source information. The received audio data and the generated data representation are utilized to subsequently train a machine learning algorithm to recognize the specific sound source during autonomous driving of the ADV in real-time.

A system and method for providing assistive listening for a plurality of listeners in an environment including a plurality of acoustic sources. A microphone array in combination with an acoustic beamforming processor configured to receive the acoustic signals within the environment and to process the acoustic signals based upon a target location of an acoustic signal selected on a listener-controlled interface device to generate a steered beam pattern. The acoustic beamforming processor further configured to transmit the steered beam pattern to the listener-controlled interface device based on the target location selected. The listener-controlled interface device configured to provide the steered beam pattern to an ear-level transducer of a hearing-impaired listener.

A method of multi-talker separation using a 3-tuple coprime microphone array, including generating, by a subarray signal processing module, a respective subarray data set for each microphone subarray of the 3-tuple coprime microphone array based, at least in part, on an input acoustic signal comprising at least one speech signal. The input acoustic signal is captured by the 3-tuple coprime microphone array. The 3-tuple coprime microphone array includes three microphone subarrays. The method includes determining, by the subarray signal processing module, a point by point product of the three subarray data sets; and determining, by the subarray signal processing module, a cube root of the point by point product to yield an acoustic signal output data. The acoustic signal output data has an output amplitude and an output phase corresponding to an input amplitude and an input phase of a selected speech signal of the at least one speech signal.

Provided are a device control method and apparatus. The method is applied to an audio device, and includes: receiving an acoustic signal set, determining a propagation characteristic of an acoustic signal in the acoustic signal set, determining, according to the propagation characteristic, a device parameter associated with audio play quality to be used by the audio device, and controlling the audio device to play audio with the device parameter.

Systems, methods, and computer-readable storage devices are disclosed for improved audio feature discovery using a neural network. One method including: receiving a trained neural network model, the trained neural network configured to output an audio feature classification of audio data; deconstructing the trained neural network model to generate at least one saliency map, the at least one saliency map providing a successful classification of the audio feature; and extracting at least one visualization of the audio feature the trained neural network model relies on for classification based on the at least one saliency map.

Provided is a voice recognition method and a voice recognition apparatus, and an air conditioner. The method includes: acquiring first voice data; adjusting, according to the first voice data, a collection state of second voice data to obtain an adjusted collection state, and acquiring the second voice data based on the adjusted collection state; and performing far-field voice recognition on the second voice data using a preset far-field voice recognition model so as to obtain semantic information corresponding to the acquired second voice data. The application can solve the problem in which far-field voice recognition performance is poor when a deep learning method or a microphone array method is used to remove reverberation and noise from far-field voice data, thereby enhancing far-field voice recognition performance.

A method and apparatus for determining a direction of infrasound. Infrasound is received by a directional infrasound sensor comprising a plurality of channels and a plurality of sensor devices. Each channel in the plurality of channels comprises a single opening at a first end of the channel and a closed end opposite the opening. The opening of each channel in the plurality of channels is pointed in a different direction from the opening of each other channel in the plurality of channels. The plurality of sensor devices includes a sensor device at the closed end of each channel in the plurality of channels. Each sensor device in the plurality of sensor devices is configured to generate a sensor signal in response to pressure. The sensor signals generated by the plurality of sensor devices are processed to determine the direction of the infrasound received by the directional infrasound sensor.

Iterative methods for direction of arrival estimation of a signal at a receiver with a plurality of spatially separated sensor elements are described. A quantized estimate of the angle of arrival is obtained from a compressive sensing solution of a set of equations. The estimate is refined in a subsequent iteration by a computed error based a quantized estimate of the direction of arrival in relation to quantization points offset from the quantization points for the first quantized estimate of the angle of arrival. The iterations converge on an estimated direction of arrival.

Example techniques relate to playback based on acoustic signals in a system including a first network device and a second network device. A first network device may detect a presence of a user using a camera and/or infrared sensors. The first network device sends, in response to detecting the presence of the user, a particular signal via the first network interface. The second network device receives data corresponding to the particular signal and plays back an audio output corresponding to the particular signal.