Abstract: Techniques for speech processing using a deep neural network (DNN) based acoustic model front-end are described. A new modeling approach directly models multi-channel audio data received from a microphone array using a first model (e.g., multi-geometry/multi-channel DNN) that includes a frequency aligned network (FAN) architecture. Thus, the first model may perform spatial filtering to generate a first feature vector by processing individual frequency bins separately, such that multiple frequency bins are not combined. The first feature vector may be used similarly to beamformed features generated by an acoustic beamformer. A second model (e.g., feature extraction DNN) processes the first feature vector and transforms it to a second feature vector having a lower dimensional representation. A third model (e.g., classification DNN) processes the second feature vector to perform acoustic unit classification and generate text data. The DNN front-end enables improved performance despite a reduction in microphones.

Abstract: Methods, apparatus, systems and articles of manufacture to reduce signature matching uncertainty in media monitoring systems are disclosed. Disclosed example apparatus are to compare first groups of monitored signatures representative of monitored media with second groups of reference signatures representative of respective reference media to determine a first sequence of match results for an observation period, the first and second groups to have a first size corresponding to a first number of signatures. Disclosed example apparatus are also to compare, in response to a discontinuity in the first sequence of signature match results, third groups of monitored signatures from the sequence of monitored signatures with fourth groups of reference signatures from the respective sequences of reference signatures to determine a second sequence of signature match results for the observation period, the third groups and the fourth groups to have a second size greater than the first size.

Abstract: A system configured to perform low input-output latency noise reduction in a frequency domain is provided. The real-time noise reduction algorithm performs frame by frame processing of a single-channel noisy acoustic signal to estimate a gain function. Accurate noise power estimates are achieved with the help of minimum statistics approach followed by a voice activity detector. The noise power and gain values are smoothed to remove any external artifacts and avoid background noise modulations. The gain values for individual frequency bands are weighted and smoothed to reduce distortion. To obtain distortionless output speech, the system performs curve fitting by separating the frequency bands into multiple groups and applying a Savitzky-Golay filter to each group. The final gain values generated by these filters are multiplied with the noisy speech signal to obtain a clean speech signal.

Abstract: Method, apparatus, and computer-readable media for touch and speech interface, with audio location, includes structure and/or function whereby at least one processor: (i) receives a touch input from a touch device; (ii) establishes a touch-speech time window; (iii) receives a speech input from a speech device; (iii) determines whether the speech input is present in a global dictionary; (iv) determines a location of a sound source from the speech device; (v) determines whether the touch input and the location of the speech input are both within a same region; (vi) if the speech input is in the dictionary, determines whether the speech input has been received within the window; and (vii) if the speech input has been received within the window, and the touch input and the speech input are both within the same region, activates an action corresponding to both the touch input and the speech input.

Abstract: A home appliance including a communication device configured to communicate with another home appliance, a microphone configured to receive a voice from a user, and a processor configured to perform signal processing on first voice data obtained from the microphone and perform voice recognition using the signal-processed first voice data. Wherein the processor generates noise data using second voice data received from the other home appliance and performs the signal processing on the first voice data using the generated noise data.

Abstract: An information processing apparatus including an acquisition section that acquires acquire an acoustic signal indicating a sound collection result of a sound collection device group and a control section that calculates a first weight in accordance with a degree to which the acoustic signal acquired by the acquisition section is a signal obtained by observing stationary noise and to apply the first weight to a noise spatial correlation matrix that is a spatial correlation matrix obtained from a noise signal.

Abstract: Sound signal processing apparatuses and methods of operating the same are provided. The sound signal processing apparatus includes: a band separator configured to separate sound signals into frequency bands; an adder configured to add sound signals; and a signal processor that is arranged between the band separator and the adder and comprises a plurality of signal processing blocks. The band separator includes elements for separating the sound signals into frequency bands, and the elements correspond one to one to the signal processing blocks.

Abstract: Methods and apparatus for digital signal processing of signals received from sensors are provided. A first input signal and a second input signal are received. A noise correlation statistic between the first input signal and the second input signal is estimated. An inter sensor signal model representative of a relationship between desired signal components present in the first input signal and the second input signal is estimated. Responsive to the noise correlation statistic meeting a predefined condition, estimating the inter sensor signal model is based on the noise correlation statistic. Responsive to the noise correlation statistic not meeting the predefined condition, estimating the inter sensor signal model is based on a constrained noise correlation statistic derived from the noise correlation statistic.

Abstract: A speech recognition method causes a computer to execute a speech recognition process. The process includes: extracting features of noise or unwanted sound included in speech data to be recognized; performing, for the speech data, a suppression process for each of the extracted features of noise or unwanted sound among suppression processes for respective features of noise or unwanted sound; and performing speech recognition for the speech data after the suppression process is performed.

Abstract: Personal audio systems and methods are disclosed. A personal audio system includes a processor to generate a personal audio stream by processing an ambient audio stream in accordance with an active processing parameter set, a circular buffer memory to store a most recent snippet of the ambient audio stream, and an event detector to detect a trigger event. In response to detection of the trigger event, a controller may extract audio feature data from the most recent snippet of the ambient audio stream and transmit the audio feature data and associated metadata to a knowledgebase remote from the personal audio system.

Abstract: Audio distortion compensation methods to improve accuracy and efficiency of audio content identification are described. The method is also applicable to speech recognition. Methods to detect the interference from speakers and sources, and distortion to audio from environment and devices are discussed. Additional methods to detect distortion to the content after performing search and correlation are illustrated. The causes of actual distortion at each client are measured and registered and learnt to generate rules for determining likely distortion and interference sources. The learnt rules are applied at the client, and likely distortions that are detected are compensated or heavily distorted sections are ignored at audio level or signature and feature level based on compute resources available. Further methods to subtract the likely distortions in the query at both audio level and after processing at signature and feature level are described.

Abstract: In methods and apparatus for performing temporal noise shaping, an apparatus may have a temporal noise shaping, TNS, tool for performing linear prediction, LP, filtering on an information signal including a plurality of frames; and a controller configured to control the TNS tool so that the TNS tool performs LP filtering with: a first filter whose impulse response has a higher energy; and a second filter whose impulse response has a lower energy than the first filter, wherein the second filter is not an identity filter, wherein the controller is configured to choose between filtering with the first filter, and filtering with the second filter on the basis of a frame metrics.

Abstract: A media system, receives a received sequence of media content, for presentation at the media system and generates a comparison fingerprint of the received sequence of media content. The comparison fingerprint is for comparison with a plurality of reference fingerprints so as to identify the received sequence of media content. The media system sends a request for identification of additional content to a server system. The request is based at least in part on the comparison fingerprint. The media system receives a response to the request, including information enabling additional content to be selected for display at the media system based at least in part on the identification of the received sequence of media content, and presents a displayed sequence of media content that includes at least a portion of the received sequence of media content and at least a portion of the additional content.

Abstract: A device includes an encoder configured to determine, during a first period, that a first audio signal is a leading signal and that a second audio signal is a lagging signal. The encoder is also configured to generate a first frame of at least one encoded signal based on a first modified version of the second audio signal that is generated by adjusting the second audio signal based on a first mismatch value. The encoder is configured to determine, during a second period, that the first audio signal is the leading signal and that the second audio signal is the lagging signal. The encoder is configured to generate a second frame of the at least one encoded signal based on a second modified version of the second audio signal that is generated by adjusting the second audio signal based on the first mismatch value and a second mismatch value.

Abstract: In methods and apparatus for performing temporal noise shaping, an apparatus may have a temporal noise shaping, TNS, tool for performing linear prediction, LP, filtering on an information signal including a plurality of frames; and a controller configured to control the TNS tool so that the TNS tool performs LP filtering with: a first filter whose impulse response has a higher energy; and a second filter whose impulse response has a lower energy than the first filter, wherein the second filter is not an identity filter, wherein the controller is configured to choose between filtering with the first filter, and filtering with the second filter on the basis of a frame metrics.

Abstract: A noise processing apparatus and a noise processing method capable of increasing cost-performance in the case of performing a noise reducing process for each band using a multiresolution transform and adding a desired noise are provided. The noise processing apparatus includes a multiresolution transform unit (32) that generates a plurality of band images in different resolutions by performing a multiresolution transform process including image size reduction on an original image, a noise reducing unit (34) that performs a noise reducing process on the band image, and a reverse multiresolution transform unit (38) that generates an output image in the same resolution as the original image by performing a reverse multiresolution transform process including image size expansion and a noise addition process on the band image subjected to the noise reducing process, the reverse multiresolution transform unit (38) performing the noise addition process in a lower resolution than the original image.

Abstract: Some embodiments of the invention are directed to techniques for suppressing or attenuating noise in audio circuitry, such as which may result from a source of electromagnetic energy being coupled to the audio circuitry. Some embodiments may employ two separate channels, a first of which captures audio information and noise, and a second of which captures the noise. Noise compensation may be performed by transforming the signal of the first channel in accordance with noise in the second channel which is identified as comprising a particular type of noise.

Abstract: Systems and methods for noise reduction are provided including operations for noisy speech signals, such as speech signals that are subject to speech processing, speech recognition and speech transmission for voice communication purposes. In one embodiment, a system for noise suppression includes an input smoothing filter to smooth magnitudes of the input spectrum, a desired noise shape determination block configured to determine a desired noise shape of the noise spectrum dependent on the smoothed-magnitude input spectrum, and a suppression factors determination block configured to determine a set of suppression factors based on the desired noise shape and the smoothed-magnitude input spectrum. In one embodiment, a filter coefficient determination block is configured to determine noise suppression filter coefficients from the desired noise shape of the noise spectrum. Embodiments are also directed to systems and methods for noise reduction.

Abstract: A controller for an audio system. The audio system comprising an audio processor and an amplifier. The controller is configured to: receive an amplifier-operating-condition-signal representative of an operating condition of the amplifier; receive a maximum-threshold-value; and generate control signaling based on the amplifier-operating-condition-signal and the maximum-threshold-value, wherein the control signaling is configured to set an operating parameter of the audio processor.

Abstract: A disclosed method includes monitoring an audio signal energy level while having a plurality of signal processing components deactivated and activating at least one signal processing component in response to a detected change in the audio signal energy level. The method may include activating and running a voice activity detector on the audio signal in response to the detected change where the voice activity detector is the at least one signal processing component. The method may further include activating and running the noise suppressor only if a noise estimator determines that noise suppression is required. The method may activate and runs a noise type classifier to determine the noise type based on information received from the noise estimator and may select a noise suppressor algorithm, from a group of available noise suppressor algorithms, where the selected noise suppressor algorithm is the most power consumption efficient.

Abstract: When conducting imaging mass analysis for a region to be measured on a sample, an individual reference value calculating part obtains a maximum value in Pi/Ii of respective measuring points, and stores the value together with measured data as an individual reference value. When performing comparison analysis for a plurality of the data obtained from different samples, a common reference value determining part reads out corresponding a plurality of the individual reference values and determines a minimum value as a common reference value Fmin. A normalization calculation processing part normalizes the respective intensity values by multiplying the intensity values read out from an external memory device by a normalization coefficient long_Max×(Fmin/Pi) obtained from the common reference value Fmin, TIC values Pi at the respective measuring points, and a maximum allowable value long_Max of a variable storing the intensity values at the time of operation.

Abstract: Sound signal processing apparatuses and methods of operating the same are provided. The sound signal processing apparatus includes: a band separator configured to separate sound signals into frequency bands; an adder configured to add sound signals; and a signal processor that is arranged between the band separator and the adder and comprises a plurality of signal processing blocks. The band separator includes elements for separating the sound signals into frequency bands, and the elements correspond one to one to the signal processing blocks.

Abstract: The present disclosure relates to processing a plurality of audio signals. The method includes receiving the plurality of audio signals in the frequency domain and determining an overall attenuation multiplier based on the plurality of audio signals and an overall lookup table that relates decibel values to different overall attenuation multipliers. The method further includes determining an attenuation vector comprising a plurality of bin-specific attenuation multipliers, each bin-specific attenuation multiplier respectively corresponding to a different frequency bin of the plurality of frequency bins. The method further includes scaling each bin-specific attenuation value in the attenuation vector with the overall attenuation multiplier, and editing each of the audio signals based on the scaled bin-specific attenuation values in the attenuation vector.

Abstract: A sound source separating device includes: a signal acquiring unit that acquires the sound signal including mixed sounds from a plurality of sound sources; a start information acquiring unit that acquires start information representing a start timing of at least one sound source among the plurality of sound sources; and a sound source separating unit that separates a specific sound source from the sound signal by setting a binary mask controlling presence of the sound source using a variable of “0” and “1” and using a Markov chain for the activation on the basis of the start information and decomposing the spectrogram generated from the sound signal into the base spectrum and the activation through non-negative matrix factorization using the set binary mask S.

Abstract: A system and method for adaptive beamforming, which is configured to process at least two input signals and to provide an output signal, wherein a first input signal of the at least two input signals includes a desired signal as a main component and a second input signal of the at least two input signals, include an undesired signal as main component, includes adaptive error processing the second input signal and at least one of the first input signal and the output signal to provide an estimated undesired signal representative of an estimate of undesired signal components included in the first input signal, and taking the difference between the estimated undesired signal and the first input signal to provide the output signal.

Abstract: Apparatuses and methods for transforming a signal from a vibration sensor into an approximation of a signal from a microphone are disclosed. The vibration sensor signal may be transformed by first obtaining a smooth average spectrum and corresponding deviation data for the vibration sensor signal. The smooth average spectrum may be passed through a mapping neural network to obtain a transformed smooth average spectrum that approximates a smooth average spectrum that would be obtained from a microphone. The transformed signal is then obtained by recombining the transformed smooth average spectrum with the deviation data.

Abstract: A system configured to improve noise cancellation by using portions of multiple reference signals instead of using a complete reference signal. The system divides a frequency spectrum into frequency bands and selects a single reference signal from a group of potential reference signals for every frequency band. For example, a first reference signal is selected for a first frequency band while a second reference signal is selected for a second frequency band. The system may generate a combined reference signal using portions of each of the selected reference signals, such as a portion of the first reference signal corresponding to the first frequency band and a portion of the second reference signal corresponding to the second frequency band. Additionally or alternatively, the system may perform noise cancellation using each of the selected reference signals and filter the outputs based on the corresponding frequency band to generate combined audio output data.

Abstract: Provided is an information providing apparatus (10) including a reliability determination unit (11) that determines a reliability of an inference result obtained by inferring an operation state of a monitored electric device group, on the basis of time-series measurement data and training data, the time-series measurement data including any one of current consumption and power consumption of the monitored electric device group and the training data associating each of a plurality of the operation states which can be assumed by the electric device group with a feature value of each of the plurality of operation states, and an output unit (12) that outputs information related to the reliability.

Abstract: Computer-based access security includes executing a system-level boot process, by a processor configured with a computing device, as a function of at least boot code, and storing, as a function of at least one operating system kernel, a value representing a time at or near completion of the system-level boot process. The security further includes receiving, as a function of a capacitive touch screen, an electrical charge operable to cause a measurable drop in voltage. Thereafter, the security includes generating, as a function of the measured voltage drop, an image. Moreover, the security includes storing the image to processor readable media configured with the computing device, wherein storing the image includes: determining a value representing an amount of time calculated from the time of completion of the system-level boot process; and associating the value representing the amount of time with the image.

Abstract: Methods and apparatus for configuring a portable communication device are disclosed. The methods include receiving an incoming audio signal, adjusting the incoming signal audio level based on a volume control setting, generating an ultrasonic audio signal having an audio level independent of the volume control setting, mixing the incoming signal with the ultrasonic signal to generate a mixed audio signal, outputting the mixed signal to a remote audio device, receiving a remote audio signal from the remote device, and comparing the remote signal audio level to one or more thresholds to determine whether to output a compensated audio signal to a speaker. The remote signal may be detected by the remote device in response to sound output from the remote device based on the mixed signal, and the compensated signal audio signal may be adjusted based on an estimated loss in loudness associated with the remote signal audio level.

Abstract: Embodiments of the present disclosure provide a method and a device for extracting a speech feature based on artificial intelligence. The method includes performing a spectrum analysis on a speech to be recognized, to obtain a spectrum program of the speech; and extracting features of the spectrum program by using an Inception convolution structure of an image recognition algorithm, to obtain the speech feature of the speech. In embodiments, by performing the spectrum analysis on the speech to be recognized, the consecutive speech to be recognized is converted into the spectrum diagram. As the Inception convolution structure is an effective image recognition manner being able to accurately recognize features of an image, the spectrum program is recognized with the Inception convolution structure to extract the relative accurate speech feature from the speech to be recognized. Thus, the accuracy rate of the speech recognition is improved.

Abstract: Systems and methods for enhancing a headset user's own voice include an outside microphone, an inside microphone, audio input components operable to receive a plurality of time-domain microphone signals, including an outside microphone signal from the outside microphone and an inside microphone signal from the inside microphone, a subband decomposition module operable to transform the time-domain microphone signals to frequency domain subband signals, a voice activity detector operable to detect speech presence and absence in the subband signals, a speech extraction module operable to predict a clean speech signal in each of the inside microphone signal and the outside microphone signal, and cancel audio sources other than a headset user's own voice by combining the predicted clean speech signal from the inside microphone signal and the predicted clean speech signal from the outside microphone signal, and a postfiltering module operable to reduce residual noise.

Abstract: A signal processing unit calculates a first difference in time of arrival of sound from a sound source to a first and to a second microphone comprising a microphone array and calculates a second difference in time of arrival, which is the difference between the first difference in time of arrival and an actual time, of arrival, and determines the position of the sound source based on the sum of the first difference in time of arrival and the second difference in time of arrival.

Abstract: Methods and circuits for reducing power dissipation in wireless transceivers and other electronic circuits and systems. Embodiments of the present invention use bias current reduction, impedance scaling, and gain changes either separately or in combination to reduce power dissipation. For example, bias currents are reduced in response to a need for reduced signal handling capability, impedances are scaled thus reducing required drive and other bias currents in response to a strong received signal, or gain is increased and impedances are scaled in response to a low received signal in the presence of no or weak interfering signals.

Abstract: Wind noise reduction is provided by obtaining a first signal from a first microphone and a contemporaneous second signal from a second microphone. A level of the first signal is compared to a level of the second signal, within a short or substantially instantaneous time frame. If the level of the first signal exceeds the level of the second signal by greater than a predefined difference threshold, a suppression is applied to the first signal.

Abstract: An apparatus and method for reducing background noise captured by a UAV acoustic sensor are disclosed. The background noise may be reduced by incorporating a known UAV acoustic signature corresponding to a determined flight parameter into an adaptive filter coupled to the acoustic sensor.

Abstract: Apparatus, methods and programs for estimating a parameter for modeling a time series signal with stochastic dynamics model, A method for obtaining data representing the mixed signals of two or more time series signals; estimating a parameter for modeling a time series signal with stochastic dynamics model. The present technology relates to a data processing apparatus, a data processing method, and a program.

Abstract: The invention relates to methods and apparatus for adjusting a level of an audio signal. An audio signal is divided into a plurality of frequency bands. Modification parameters are obtained for at least one of the plurality of frequency band. Gain factors are derived for at least one of the plurality of frequency bands, the gain factors determined based on the amplitude scale factors. The gain factors are smoothed. A level of noise from noise compensation factors is determined. The gain factors are applied to at least one of the frequency bands to generate gain adjusted frequency bands. The level of noise is adjusted based on the gain adjusted frequency bands. At least one of the frequency bands is filtered with a filter generated with the filter coefficients. The plurality of frequency bands is synthesized to generate an output audio signal.

Abstract: A method for controlling echo of a voice recognition system includes receiving at least two audio input signals corresponding to sound sensed by at least two microphones in a physical space. A first audio input signal of the at least two audio input signals is received on a primary channel, and each remaining audio input signal is received through a respective secondary channel. The method includes selecting, by a processor, based on an echo power level of a speaker in the physical space, a subset of echo control functions (ECFs) from among a plurality of ECFs of a multistage echo control system. Each ECF modifies the at least two audio input signals to reduce echo. The method includes generating a corresponding number of audio output signals by processing the signals received on the primary and secondary channels through the selected subset of ECFs, and outputting the audio output signals.

Abstract: Disclosed methods and apparatuses determine an inter-channel time difference of a multi-channel audio signal by determining a set of local maxima of a cross-correlation function involving at least two different channels of the multi-channel audio signal, for positive and negative time-lags and associating each local maximum with a corresponding time-lag. Local maximums for the positive and negative time-lags are selected from the set as respective positive and negative time-lag inter-channel correlation candidates. Processing also includes evaluating whether there is an energy-dominant channel, when the absolute value of a difference in amplitude between the inter-channel correlation candidates is smaller than a first threshold, and, when there is an energy-dominant-channel, identifying the sign of the inter-channel time difference and extracting a current value of the inter-channel time difference based on the time-lag corresponding to the positive or negative time-lag inter-channel correlation candidate.

Abstract: Systems and methods for enhancing reverberated audio signals are disclosed. In one embodiment, a method is disclosed comprising receiving an audio signal; partitioning a frequency domain representation of the audio signal into a plurality of sub-band vectors; inputting each sub-band vector into a corresponding deep neural network; calculating, using the corresponding deep neural networks, a plurality of output vectors for each sub-band; concatenating the plurality of output vectors to generate a clean audio feature matrix; and converting the clean audio feature matrix into a time-domain audio signal.

Abstract: A device (160) and a method are proposed for tuning a frequency-dependent attenuation stage (122), with the purpose of suppressing non-linear distortion occurring in an audio reproduction system (120) associated with the frequency-dependent attenuation stage. The device comprises a receiving section (162) adapted to receive data representing an output acoustic signal (140) from the audio reproduction system, recorded upon excitation of the audio reproduction system by a predetermined input data signal (110)a first distortion detection section (163) adapted to detect presence of non-linear distortion based on the received data and to apply psycho-acoustic compensation to the detected non-linear distortion and a control section (164) adapted to determine, based on the psycho-acoustically compensated non-linear distortion, control information (170) suitable for controlling the frequency-dependent attenuation stage.

Abstract: Devices and methods for howling suppression. One method includes receiving, via a microphone, an acoustic signal from a communication device operating in an acoustic field with the microphone. The method includes determining a reflection pattern for the acoustic field based on the acoustic signal, and determining an acoustic characteristic for the acoustic field based on the reflection pattern. The method includes determining, based on the acoustic characteristic, a plurality of howling zones for the acoustic field, each zone defined by first and second proximity thresholds. The method includes, for each of the howling zones, determining an attenuation level for the zone based on the proximity thresholds and the acoustic characteristic. The method includes determining a distance between the microphone and the communication device, selecting one of the howling zones based on the distance, and adjusting a volume of a loudspeaker based on the attenuation level for the selected howling zone.

Abstract: Information from microphone signals from a microphone array may be used to identify persistent sources, such as televisions, radios, washing machines, or other stationary sources. Values representative of broadside conditions for each pair of microphone signals are received from the microphone array. By monitoring broadside conditions for microphone pairs, a position of a sound source may be identified. If a sound source is frequently identified with a broadside of the same microphone pair, then that sound source may be identified as a persistent noise source. When a broadside of a pair of microphones is identified with a noise source, a beamformer may be configured to decrease contribution of that pair of microphones to an audio signal formed from the microphone array.

Abstract: The present invention proposes a method for providing at least one indication regarding whether an input signal contains a tonal signal component. The proposed method comprises decomposing the input signal into a plurality of band-limited input signals and then for each of one or more of the plurality of band-limited input signals performing the following step: estimating parameters of a parametric tonal signal model adapted to model and predict a tonal-only signal component of a selected one of the one or more of the plurality of band-limited input signals; synthesizing based on the estimated parameters a corresponding predicted band-limited input signal; and determining a measure of mismatch between the selected band-limited input signal and the corresponding predicted band-limited input signal. Subsequently, the at least one indication is provided based on the measure of mismatch determined for at least one of the one or more of the plurality of band-limited input signals.

Abstract: A system for self-organized acoustic signal cancellation over a network is disclosed. The system may transmit an acoustic sounding signal to an interfering device so that a channel measurement may be performed for a channel between the interfering device and an interferee device. The system may receive the channel measurement for the channel from the interfering device and also receive a digitized version of an audio interference signal associated with the interfering device. Based on the channel measurement and the digital version of the interference signal, the system may calculate a cancellation signal prior to the arrival of the original over-the-air audio interference signal that corresponds to the digital version of audio interference signal. The system may then apply the cancellation signal to an audio signal associated with the interferee device to remove the interference signal from the audio signal.

Abstract: One method of operation includes beamforming a plurality of microphone outputs to obtain a plurality of virtual microphone audio channels. Each virtual microphone audio channel corresponds to a beamform. The virtual microphone audio channels include at least one voice channel and at least one noise channel. The method includes performing voice activity detection on the at least one voice channel and adjusting a corresponding voice beamform until voice activity detection indicates that voice is present on the at least one voice channel. Another method beamforms the plurality of microphone outputs to obtain a plurality of virtual microphone audio channels, where each virtual microphone audio channel corresponds to a beamform, and with at least one voice channel and at least one noise channel. The method performs voice recognition on the at least one voice channel and adjusts the corresponding voice beamform to improve a voice recognition confidence metric.

Abstract: A noise detector and a sound signal output device are provided that can detect high accuracy various types of noise including composite noise. The noise detector (30) for detecting noise contained in an input signal (s1) includes a noise discrimination portion having a plurality of discrimination portions to which an input signal is input and a noise determination portion (35) determining noise based on individual discrimination results of the plurality of discrimination portions.

Abstract: A spatial adaptation system for multiple-microphone sound capture systems and methods thereof are described. A spatial adaptation system includes an inference and weight module configured to receive a inputs. The inputs based on two or more input signals captured by at least two microphones. The inference and weight module to determine one or more weight values base on at least one of the inputs. The spatial adaptation system also including a noise magnitude ratio update module coupled with the inference and weight module. The noise magnitude ratio update module to determine an updated noise target based on the one or more weight values from the inference and weight module.

Abstract: A system comprising at least one processor; and at least one storage device. The storage device(s) store instructions that, when executed, cause the at least one processor to: prior to enabling output of an audio signal based on an audio data stream, detect, within the audio data stream, an indication of a target sound that corresponds to one of a plurality of sounds that are expected to cause distraction, replace, within the audio data stream, the indication of the target sound with an indication of a replacement sound, wherein the replacement sound is a less distracting version of the target sound, and after replacing the indication of the target sound with the indication of the replacement sound, output the audio data stream.