88 results about "Non stationary noise" patented technology

A two microphone noise reduction system is described. In an embodiment, input signals from each of the microphones are divided into subbands and each subband is then filtered independently to separate noise and desired signals and to suppress non-stationary and stationary noise. Filtering methods used include adaptive decorrelation filtering. A post-processing module using adaptive noise cancellation like filtering algorithms may be used to further suppress stationary and non-stationary noise in the output signals from the adaptive decorrelation filtering and a single microphone noise reduction algorithm may be used to further provide optimal stationary noise reduction performance of the system.

A noise reduction device is configured by use of: means for calculating a predetermined constant, and a predetermined reference signal Rω(T) in the frequency domain, respectively by use of adaptive coefficients Wω(m), and for thereby obtaining estimated values Nω and Qω(T) respectively of stationary noise components, and non-stationary noise components corresponding to the reference signal, which are included in a predetermined observed signal Xω(T) in the frequency domain; means and for applying a noise reduction process to the observed signal on the basis of each of the estimated values, and for updating each of the adaptive coefficients on the basis of a result of the process; and an adaptive learning means and for repeating the obtaining of the estimated values and the updating of the adaptive coefficients, and for thereby learning each of the adaptive coefficients.

The invention discloses a speech enhancement method using a speech existence probability, after the processing of the invention, the speech quality is higher, and the noise reduction amount is larger.The method is realized by the following technical scheme that: on the basis of a MIS measure speech enhancement method, the speech existence probability is used, an inputted noisy speech is sampled,framed and windowed, and then a noisy speech spectrum is obtained through fast Fourier transform (FFT); the noise estimation of the obtained speech spectrum is carried out, based on a non-stationary noise minimum search algorithm of statistical information, the smoothing between front and back frame noise estimation values is carried out by using inter-frame correlation, and a noise power spectrumis estimated; at the same time, speech prior signal-to-noise ratio estimated values obtained in several front and back frames are smoothened; then the speech existence probability and a MIS measure gain factor are combined, the obtained noisy speech spectrum is multiplied by a new gain factor, and a spectrum of the enhanced speech is obtained; and the inverse fast Fourier transform (IFFT) is performed to obtain an enhanced time domain speech signal.

The invention discloses a self-adaptation voice detection method based on multi-domain joint estimation. Voice can be detected in a complex environment in a self-adaptation mode, accuracy and reliability are good, and the complexity of detection is low. The voice detection method includes the steps of signal preprocessing, frequency spectrum stability judging, stable noise voice detecting, non-stable noise voice detecting and voice segment signal outputting.

The invention relates to an active noise reduction method for an automobile and belongs to the voice signal processing technical field. The active noise reduction method guides the secondary sound source and adopts the self-adaptive algorithm for controlling the sound signal sent by the secondary sound source, and the outputted secondary sound wave after self-adaptive convergence at the noise reducing point has constant amplitude and reversal phase with the noise at the point for getting the noise reducing target. The algorithm and the method structure are improved based on the original active noise reduction method, the recursive least square method is used for replacing the least mean square algorithm as the core self-adaptive algorithm for the main path transfer function estimation and the secondary path transfer function estimation; the method has strong capability of eliminating the pulse noise and non-stationary noise and has good noise reducing error and noise reducing feed, the stability problem caused by the signal dependency is improved for the retaining signal in the automobile, the useful signal is kept while the noise is reduced and the signal to noise ratio is greatly improved.

An electronic device for suppressing noise in an audio signal is described. The electronic device includes a processor and instructions stored in memory. The electronic device receives an input audio signal and computes an overall noise estimate based on a stationary noise estimate, a non-stationary noise estimate and an excess noise estimate. The electronic device also computes an adaptive factor based on an input Signal-to-Noise Ratio (SNR) and one or more SNR limits. A set of gains is also computed using a spectral expansion gain function. The spectral expansion gain function is based on the overall noise estimate and the adaptive factor. The electronic device also applies the set of gains to the input audio signal to produce a noise-suppressed audio signal and provides the noise-suppressed audio signal.

The invention belongs to the field of video monitoring, and provides a speech endpoint detecting method and device. The method comprises the following steps: sampling data of an input speech signal, and preprocessing the sampled speech signal; adding a Hamming window to the preprocessed speech signal for framing and recording as Rn (n is more than 0 and less than or equal to N), wherein N is the total number of frames; calculating a frequency spectrum information entropy of the n-th speech signal; and determining the frame as a speech frame if the frequency spectrum information entropy of the n-th speech signal is more than a set threshold value, and otherwise, determining the frame as a non-speech frame. The method applies the frequency spectrum entropy as a characteristic for distinguishing a speech frame from a non-speech frame, can effectively distinguish speech frames from non-speech frames, and has a good detection effect for low signal to noise ratio environments, so the defects that the traditional frequency spectrum entropy-based algorithm only considers the frequency spectrum information of the current frame, the noise frequency spectrum information entropy greatly fluctuates in a non-stationary noise environment, and the difficulty of threshold value selection is increased can be overcome.

This invention relates to a method for testing directions of coherent signal sources under non-stable noise background including: collecting data received by channels and storing them in an EMS memory of the system, making adaptive uniform process to data received by the channels to form an array covariance matrix, utilizing a forward space smooth technology to preprocess it smoothly to constitute a forward space smooth covariance matrix and utilizing a forward and backward space smooth technology to smoothly pre-process the matrix to constitute a generalized covariance matrix based on the covariance matrix and a conjugated matrix and carrying out spectrum estimation to the generalized matrix with a sub-space class algorithm to identify a real incident direction of image false peak and signal source and output the real direction of the estimated signal source.

A In a noise suppression apparatus, an extractor extracts a noise component from an audio signal. A stationary noise estimator estimates stationary noise included in the noise component. A first noise suppressor removes a spectrum of the stationary noise from a spectrum of the audio signal to an extent determined by a subtraction factor. A non-stationary noise estimator estimates a spectrum of non-stationary noise by subtracting the spectrum of the stationary noise from the spectrum of the noise component. A factor setter generates a filtering factor for emphasizing a target sound component contained in the audio signal from the spectrum of the non-stationary noise. A second noise suppressor performs a filtering process using the filtering factor on the audio signal after processing of the first noise suppressor. An index calculator calculates a kurtosis change index representing an extent of change of kurtosis in a frequence distribution of magnitude of the audio signal between the kurtosis observed when processing of the first noise suppression part is performed and the kurtosis observed when processing of the second noise suppression part is performed. A factor adjuster variably controls the subtraction factor according to the kurtosis change index.

The invention discloses a single-channel speech enhancement method based on an attention-gated recurrent neural network. The method comprises the steps of framing and winding single-channel speech with noise, and extracting 38-dimensional signal features; constructing a deep recurrent neural network for single-channel speech enhancement; constructing a training data set by using a pure speech library and a noise library; training the constructed deep recurrent neural network; inputting the extracted speech features with noise into the trained deep recurrent neural network, outputting a frequency band gain estimation value of the speech with noise, smoothing and interpolating to obtain interpolation gain; enabling the interpolation gain to act on the single-channel speech with noise so as to obtain an enhanced speech spectrum. The single-channel speech enhancement method can effectively inhibit the noise including nonstationary noise, and maintains low enough computation complexity, thus being used for real-time single-channel speech enhancement; the method is ingenious and novel in conception, thus having a good application prospect.

The invention discloses a self-adaptive spectral subtraction real-time speech enhancement method. The method comprises: establishing a dynamic threshold including discrimination of voice and no voice of noise voice, providing noise spectrum time varying update principles according to the dynamic threshold; making full use of correlation extraction information among adjacent frames, and realizing a pure voice spectrum smooth estimation method. Aimed at a practical problem that voice signals are difficult to extract in unstable noise and strong background noise, the self-adaptive spectral subtraction speech enhancement method is provided. The method uses a rapid tracking noise algorithm to perform smooth update on the unstable noise frame by frame, and can preferably estimate noise spectrums. The algorithm can effectively restrain background noise, and improves voice quality and intelligibility after noise reduction. The method is low in calculation cost, and is easy to realize, and has good real-time property. The method provides a new approach for denoising of strong background noise and detection of weak signals.

The invention provides a speech separation method based on a fuzzy membership function, and belongs to speech separation methods. The fuzzy membership function is combined in the speech separation method, so that more accurate definition of a membership degree of speech time frequency units to a target signal is obtained. An auditory oscillation model is built through human ear auditory system simulation, and speech pitch characteristics are extracted. The speech time frequency units are marked according to pitch cycle characteristics to form foreground streams and background streams. Whether the corresponding time frequency units are targets or noise is judged according to different marks. In the synthesis stage, a target unit multiplies a high weight, a noise unit multiplies a low weight, and resynthetized speech is obtained. By means of the speech separation method, the pitch cycle can be estimated more precisely, the time frequency units can be marked more accurately on the basis of characteristic clues, and the more complete target speech can be obtained. Due to the fact that the method is based on the pitch characteristics of the speech, good separation effects in complex and non-stationary noise are achieved, and the application range is wide.

A digital electronic stethoscope includes an acoustic sensor assembly that includes a body sensor portion and an ambient sensor portion, the body sensor portion being configured to make acoustically coupled contact with a subject while the ambient sensor portion is configured to face away from the body sensor portion so as to capture environmental noise proximate the body sensor portion; a signal processor and data storage system configured to communicate with the acoustic sensor assembly so as to receive detection signals therefrom, the detection signals including an auscultation signal comprising body target sound and a noise signal; and an output device configured to communicate with the signal processor and data storage system to provide at least one of an output signal or information derived from the output signal. The signal processor and data storage system includes a noise reduction system that removes both stationary noise and non-stationary noise from the detection signal to provide a clean auscultation signal substantially free of distortions. The signal processor and data storage system further includes an auscultation sound classification system configured to receive the clean auscultation signal and provide a classification thereof as at least one of a normal breath sound or an abnormal breath sound.

The invention discloses a speech enhancement method based on Gaussian mixture model (GMM) noise estimation, wherein the GMM is used for estimating background noise and a spectral subtraction coefficient, spectral subtraction is conducted on noisy speech, and pure speech is recovered. Firstly, the noisy speech is preprocessed so as to obtain the amplitude and phase of the noisy speech, the amplitude is used for noise estimation and spectral subtraction, and the phase is used for recovering a time-domain signal; then, the GMM is used for estimating noise parameters and pure speech cepstrum characteristics from the noisy speech in real time, and the spectral subtraction coefficient is calculated according to the estimated pure speech cepstrum characteristics; finally, spectral subtraction is conducted on the frequency spectrum of the noisy speech, the time-domain signal is recovered, and enhanced speech is obtained according to an overlap-add method. According to the speech enhancement method, the capability of the speech enhancement algorithm to track non-stationary noise can be improved remarkably.

By using the adaptive filter, the reference input signal is processed so as to identify a pseudo-signal of a particular signal to be deleted. The pseudo-signal is subtracted from the mixture containing a target signal inputted from a microphone, the particular signal to be deleted, and a noise so as to obtain an error signal. A stationary noise is estimated to obtain a stationary noise estimated value. A non-stationary noise is estimated to obtain a non-stationary noise estimated value. The stationary noise estimated value is mixed with the non-stationary estimated value to obtain a mixed noise estimated value. An update amount is calculated according to a correlation value between the error signal and the reference input signal, and the mixed noise estimated value. According to the update amount, a coefficient of the adaptive filter is updated.

A system and method to estimate a location of a speaker who produces a sound signal even in a non-stationary noise environment. The system includes a signal input module receiving a first sound signal from an outside; an initialization module preparing a sound map, on which a spatial spectrum for the first sound signal, produced from at least one fixed sound source and received by the signal input module, is arranged, and estimating a location of the fixed sound source; a storage module storing information about the estimated location of the fixed sound source; and a speaker's location estimation module estimating a location where a second sound signal is produced using information about the spatial spectrum for sound signals including the first sound signal received by the signal input module and the information about the estimated location of the fixed sound source.

The invention relates to a speech enhancement method of a microphone array under a non-stationary noise environment. The probability of an effective signal is accurately judged by means of phase information between passageways in order to improve non-stationary noise environment interference resistant performance. The method which estimates the probability of the effective signal specifically comprises the following steps: (51) according to a required beam width, calculating a threshold T1=NL sin (theta) / Fs, wherein N represents frame length, L represents microphone unit distance, theta represents beam width, and Fs represents sample frequency, (52) calculating the valve of the sum of phase differences omega of input signals of all passageways on all frequency points pd (n, omega); and (53) when the pd (n, omega)>omega (M-1) T1, the effective signal probability exists on the frequency point p1 (n, omega)=1, otherwise, p1 (n, omega)=0, wherein M represents the number of microphones. Due to the fact that the phase information between the passageways is used, judgment to the probability of the effective signal is enabled to be more accurate, the signal to noise ratio of picked up voice is improved, whole performance is improved, and particularly non-stationary noise environment interference resistant performance is improved.

A method of separating and localizing sources of traveling waves, by obtaining linearly independent time-differentiated instantaneous observations of the sources, from spatial derivatives of the traveling wave acquired using a distributed sensor or a sensor array. The sources are blindly separated by direct application of (static) independent component analysis on the time-differentiated observations, yielding both the sources and their direction cosines relative to the sensor geometry. The method is suited for arrays of small aperture, with dimensions shorter than the coherence length of the waves. In one preferred embodiment, three sources are separated and localized from differential observations on four coplanar sensors positioned on the corners of a square. Applications include miniature microphone arrays for intelligent hearing aids with adaptive suppression of interfering signals and nonstationary noise, and integrated RF arrays for resolving or localizing multiple users or channels from a base station or mobile unit.

The invention discloses a voice signal processing method and device. The voice signal processing method includes acquiring energy distribution characteristics of a voice signal frame; judging whether the voice signal frame is a noise frame or not according to the energy distribution characteristics. By the voice signal processing method and device, the problem of poor judgment effect on unstable noise fast in changing in the prior art is solved, and judgment accuracy of the noise frame in the voice signal is improved.

The invention discloses a noise robustness endpoint detection method based on a likelihood ratio test. The improvement is achieved from the three aspects of signal to noise ratio estimation, threshold value robustness setting and trailing distortion elimination respectively, so that the suggested algorithm has a better detection property under a low signal to noise ratio environment, in particular under a non-stationary noise environment compared with the prior art. The method and a multi-observation likelihood ratio test algorithm based on harmonic wave features have similar voice boundary detection accuracy, however, the method can have better voice detection precision than the multi-observation likelihood ratio test algorithm based on the harmonic wave features, and therefore it can be proved that the method is more excellent in performance than a traditional method. Meanwhile, the method has the similar performance under the 15dB and 25dB signal to noise ratio, and it shows that the method has good robustness to noise. The noise robustness endpoint detection method can be used as an important and effective method for front end preprocessing of a voice recognition system or a voiceprint recognition system in an actual environment, and thus good application value can be achieved.

The invention relates to a DNN voice enhancement model based on MEE optimization criteria, and belongs to the field of voice enhancement of artificial intelligence. The model comprises an input layer,a hidden layer and an output layer, and the whole training model is divided into a training stage and an enhancement stage. In the training stage, and pure voice is added with different types of noises to establish mixed noise included voices in different signal to noise ratios; and features are extracted from the mixed voices and input to the DNN for training. In the enhancement stage, same features are extracted from the mixed voices to be tested, and input to the trained DNN for decoding, the network outputs estimation of the features of the pure voice, and waveform reconstruction is carried out to obtain a noise-reduced voice file. The model is highly universal for noise reduction of voices including non-smooth noises in practical problems.

The present invention discloses a coherent signal source MIMO radar angle estimation method in non-stationary noise. The method concretely comprises the steps of: 1, performing system initialization;2, obtaining signal data being subjected to MIMO radar matching filter processing, and calculating a signal covariance matrix; 3, employing a conventional subspace algorithm to achieve angle estimation of an incoherent signal source; 4, constructing a new Toeplitz matrix; 5, rejecting the incoherent signal source, and obtaining a covariance matrix of the coherent signal source; 6, employing an m-Capon algorithm to perform angle estimation of the coherent signal source, and estimating a direction of arrival of a target; and 7, estimating a direction of departure of the target. The coherent signal source MIMO radar angle estimation method in non-stationary noise achieves angle estimation of the coherent signal source.

A speech enhancement method and a speech enhancement system are provided. The speech enhancement method performs two-stage noise suppression by using digital signal processing and neural network approach. The first-stage noise suppression generates artifact signals by reducing stationary noise in the digital audio signals. The second-stage noise suppression performs voice activity detection and further reduces non-stationary noise in the artifact signals. The result of the voice activity detection is fed back to establish or update a noise model used in the first-stage noise suppression.

The invention discloses a dual-microphone directional noise inhibition system and method fused with fundamental frequency information. The system comprises a main sound signal input device, an auxiliary sound signal input device, a single-channel voice signal separating module and a dual-channel voice signal separating module. The method comprises the steps that the main sound signal input device collects main voices and inputs the main voices into the single-channel voice signal separating module, and the auxiliary sound signal input device collects background noise signals; and output signals of the single-channel voice signal separating module and the auxiliary sound signal input device are all input into the dual-channel voice signal separating module, and finally processed sound signals are output. According to the invention, a contradiction problem between noise remaining and voice distortion is fully considered, noises can be effectively inhibited on the premise that voices are not distorted, and most non-stationary noises in the real environment can be processed.