331 results about "Time frequency transform" patented technology

The invention includes a method, apparatus, and computer program to selectively suppress wind noise while preserving narrow-band signals in acoustic data. Sound from one or several microphones is digitized into binary data. A time-frequency transform is applied to the data to produce a series of spectra. The spectra are analyzed to detect the presence of wind noise and narrow band signals. Wind noise is selectively suppressed while preserving the narrow band signals. The narrow band signal is interpolated through the times and frequencies when it is masked by the wind noise. A time series is then synthesized from the signal spectral estimate that can be listened to. This invention overcomes prior art limitations that require more than one microphone and an independent measurement of wind speed. Its application results in good-quality speech from data severely degraded by wind noise.

Systems and methods according to the invention employ an acceleration sensor to characterize the synchrony or dyssynchrony of the left ventricle. Patterns of acceleration related to myocardial contraction can be used to assess synchrony or dyssynchrony. Time-frequency transforms and coherence are derived from the acceleration. Information and numerical indices determined from the acceleration time frequency transforms and coherence can be used to find the optimal pacing location for cardiac resynchronization therapy. Similarly, the information can be used to optimize timing intervals including V to V and A to V timing.

This invention provides a method and a system for coding / decoding frequency domain parameter stereos and multiple tracks, which converts signals of right and left tracks to a residual signal of a sum-difference tack and a residual signal of an estimated difference track, carries out time frequency conversion to residual signals of the estimated difference track, the original difference track and the original sum track to pick up the HF component and low frequency component of the residual signal of the original difference track, which is coded in quantization to utilize the HF component to get estimated residual RF component of the right and left tracks and gets the original residual RF component of the right and left tracks to be controlled in gain to get a gain coefficient of the difference of the two tracks and coded in quantization and finally multiplexes the coded data and edge information to get a frequency domain parameter stereo bit stream.

The present invention relates to a method of signal processing of electrograms for use in medical devices, preferably by time-frequency transforms. The present invention additionally relates to a system for receiving and analyzing such signals. The present invention preferably is a method utilizing the time-frequency transforms, such as wavelet transforms, for the purpose of artifact removal from EGs. These transforms decompose a signal in both time and frequency domains, and therefore, are well suited for non-stationary signal analysis. As a result, dissimilar signal features are well localized both in time and frequency, which potentially provides a good separation between the signal of interest and artifacts. This particularly applies to large-amplitude artifacts corrupting EGs.

The invention discloses a method for intelligently diagnosing a rotating machine fault feature based on a deep CNN model. The method comprises: (1) acquiring rotating machine fault vibration signal data, segmenting the data and performing de-trend item preprocessing; (2) performing short-time Fourier time-frequency transform analysis on the signal data to obtain the time-frequency representation of each vibration signal, and displaying the time-frequency representation with a pseudo-color map; (3) reducing the image resolution by an interpolation method and superimposing respective images to form a training sample and a test sample as inputs of the CNN; (4) constructing the deep CNN model including an input layer, two convolution layers, two pooling layers, a fully connected layer, and a softmax classification layer and an output layer; and (5) introducing the training sample into the model for training, obtaining a convolution feature, a pooling feature and a neural network structuralparameter, and diagnosing unknown fault signals according to the constructed deep CNN. The method has better accuracy and stability than an existing time-domain or frequency-domain method.

A signal analysis system and method for analyzing an input signal acquired from a mechanical system. The mechanical system may include at least one rotating apparatus. The signal analysis system may be configured to: (a) receive samples of the input signal, (b) perform an invertible joint time-frequency transform (e.g. a Gabor transform) on the samples of the input signal to produce a first array of coefficients which depend on time and frequency, (c) select first coefficients from the first array which correspond to a first subset of one or more order components in the input signal, (d) generate a time domain signal from the first coefficients, and (e) present the time domain signal to a user on a presentation device. The signal analysis system generate the time domain signal from the first coefficients by performing an inverse joint time-frequency transform on the first coefficients. The signal analysis system extract the one or more order components for presentation to the user by masking out all coefficients except those corresponding to the one or more components. Conversely, the signal analysis system may suppress the one or more order components, i.e. present to the user the input signal minus the one or more order components by masking out coefficients corresponding to the one or more components and keeping the remaining coefficients.

A low-bitrate encoding system includes: a time-frequency transform unit transforming an input time-domain audio signal into a frequency-domain audio signal; a frequency component processor unit decimating frequency components in the frequency-domain audio signal; a psychoacoustic model unit modeling the received time-domain audio signal on the basis of human auditory characteristics, and calculating encoding bit allocation information; a quantizer unit quantizing the frequency-domain audio signal input from the frequency component processor unit to have a bitrate based on the encoding bit allocation information input from the psychoacoustic model unit; and a lossless encoder unit encoding the quantized audio signal losslessly, and outputting the encoded audio signal in a bitstream format. Using the low-bitrate encoding system, it is possible to effectively compress data at a low bitrate, and thus to provide a high quality audio signal.

Hierarchical audio coding and decoding method and system and hierarchical audio coding and decoding method for transient signals are provided. In the present invention, by introducing a processing method for transient signal frames in the hierarchical audio coding and decoding methods, a segmented time-frequency transform is performed on the transient signal frames, and then the frequency-domain coefficients obtained by transformation are rearranged respectively within the core layer and within the extended layer, so as to perform the same subsequent coding processes, such as bit allocation, frequency-domain coefficient coding, etc., as those on the steady-state signal frames, thus enhancing the coding efficiency of the transient signal frames and improving the quality of the hierarchical audio coding and decoding.

A signal analysis system and method for analyzing an input signal acquired from a mechanical system. The mechanical system may include at least one rotating apparatus. The signal analysis system may be configured to: (a) receive samples of the input signal, (b) perform an invertible joint time-frequency transform (e.g. a Gabor transform) on the samples of the input signal to produce a first array of coefficients which depend on time and frequency, (c) compute an instantaneous rotation frequency signal, (d) select first coefficients from the first array which correspond to one or more order components in the input signal, i.e. one or more multiples of the instantaneous rotation frequency signal, (e) generate a time domain signal from the first coefficients, and (f) present the time domain signal to a user on a presentation device. The signal analysis system may generate the time domain signal from the first coefficients by performing an inverse joint time-frequency transform on the first coefficients.

In a method of acoustic feedback suppression, digitized time-domain samples of acoustic signals are obtained, and discrete time-frequency transformation is performed on the digitized time-domain samples to generate a plurality of frequency bins of a frequency resolution. A howling frequency bin is identified, where the howling frequency has the maximum magnitude among the plurality of frequency bins. A peak frequency within the howling frequency bin is detected, and this peak frequency is suppressed.

An audio encoding apparatus that allows a decoded signal exhibiting an excellent sound quality to be obtained on a decoding side. In the audio encoding apparatus (1000A), a time-frequency transform unit (1001) uses a time-frequency transform, such as a discrete Fourier transform (DFT) or a modified discrete cosine transform (MDCT), to transform a time domain signal (S(n)) to a frequency domain signal (spectrum factor) (S(f)). A psychoacoustic model analyzing unit (1002) performs a psychoacoustic model analysis of the frequency domain signal (S(f)), thereby obtaining a masking curve. An acoustic sense weighting unit (1003) estimates, based on the masking curve, an importance degree of acoustic sense, and determines and applies the weighting factors of respective spectrum factors to the respective spectrum factors. An encoding unit (1004) encodes the frequency domain signal (S(f)) as weighted in terms of the acoustic sense. A multiplexing unit (1005) multiplexes and transmits the encoded parameters.

The invention provides a time-frequency transform based method for identifying parameters of a synchronous generator. The method includes steps of S1. carrying out Laplace transform on a differential equation of a d axle and a q axle of a generator in time domain in a generator model to obtain an equation of the d axle and the q axle of the generator in frequency domain, and carrying out inverse Laplace transform to obtain an integral equation of the d axle and the q axle of the generator in the time domain; S. replacing state variable of the integral equation of the d axle and the q axle of the generator in the time domain; S3. dispersing the integral equation of the d axle and the q axle of the generator in the time domain into that measurable column vectors or integration of the measurable column vectors are equal to a matrix multiplying by column vectors of coefficients of the integration equation, and solving to obtain the column vectors of the coefficients of the integration equation; and S4. solving to obtain parameters of the generator model according to the relation between the column vectors of the coefficients of the integration equation and parameters of the generator. A differential equation of a generator is converted to an integration equation according to the time-frequency transform based method for identifying the parameters of the synchronous generator, stability of calculating a value is facilitated, and the problem of an unstable identification result is solved.

The invention discloses a multi-channel audio signal compressing method based on tensor decomposition, and belongs to the technical field of audio signal processing, in particular to the technical field of spatial audio coding and decoding. The method comprises the following steps: overlapping and framing an audio signal of each channel and carrying out time frequency transform on each frame of signal to obtain a frequency domain coefficient; combining all channels and the frequency domain coefficients of all frame sequences to establish a three-order tensor signal; carrying out tensor decomposition on the three-order tensor signal so as to obtain a low-rank nuclear tensor for coding transmission; reconstructing a tensor signal by using the low-rank nuclear tensor combined and recovered at a decoding end and a low-rank projection matrix trained in advance; and carrying out inverse transformation and overlap-add on the reconstructed tensor signal in each channel to recover a multi-channel audio signal. The multi-channel audio signal compressing method has the advantages as follows: as the multi-channel audio signal is analyzed, coded and decoded through the combination of time frequency transform and tensor decomposition and redundant information is removed by using correlations between channels and within the channels, the compression efficiency of the multi-channel audio signal can be increased to a greater degree.

The present invention can implement the utilization of electromagnetic depth-measuring exploration method in frequency-domain method and time-domain method. Said invention utilizes that exciting and receiving a series of frequency (period) pulses to research the law of whole spectrum, at the same time, research the signal attenuation law of every frequency (period), from short period to long period, the reflected underground informations can be progressively deepened, the superposition of these informations can raise exploration accuracy, and the time-domain and frequency-domain boty can be used for making mutual check by means of time-frequency conversion so as to greatly raise exploration accuracy. Said invention can be mainly used for making exploration of oil, gas, mineral resources and geothermal resource.

The invention belongs to the technical field of radar emitter signal modulation recognition, and particularly relates to a modulation recognition method for extracting time-frequency image features byjoint entropy and pre-training CNN. The method includes the following steps: firstly, performing time-frequency transformation on 9 types of radar signal sets to be identified to obtain a time-frequency image; and then based on a pre-training convolutional neural network model imagenet-vgg-verydeep-19 provided by the MatConvNet official website, constituting an FT-VGGNet-fc6 feature migration extraction module from an input layer to an fc6 full connection layer; and then, sending an adjusted image to the feature migration extraction module, and outputting time-frequency image features of radar signals; performing graying on the adjusted image, and manually extracting the Renyi entropy of the processed image; and then, dividing a training set and a test set according to a certain proportion, and selecting the training set to train an SVM classifier; and finally, adopting the trained SVM classifier to recognize the training set of time-frequency images, and adopting a data set composedof 9 types of radar signals with multiple signal-to-noise ratios to verify a recognition rate of an FT-VGGNET-fc6-SVM classifier.

The invention discloses an audio codec implementing method, having low calculating complexity and low storage capacity and able to assure audio compression quality, and it is an audio compressing coding technique applied to handheld equipment, system on chip (SOC) or application specific integrated circuits (ASICs) products and embedded systems. At the coder end: 1. making time-frequency transform on audio signals and transforming them into frequency domain and obtaining non-quantized spectrum data; 2. according to the mentioned non-quantized spectrum data and target allowable code quantity, and obtaining corresponding optimum quantization factor information, frequency band group information, code table index information and quantized spectrum information; 3. calculating and formatting code flow; and 4. outputting formatted code flow; at the decoder end: by code flow formatted resolution, it decodes and inversely quantizes spectrum of each frame, makes frequency-time transform and reconstructs time-domain audio data and finally reconstructs time-domain signal of each channel.

The invention provides an inverse Q filtering method for a seismic wave signal. The method comprises the following steps: extracting a quality factor Q value from the seismic wave signal collected from a seismic channel; performing time-frequency transformation on the collected seismic wave signal so as to acquire a seismic wave frequency spectrum shown in the specification, wherein tau is a travel-time and omega is an angular frequency; and performing inverse Q filtering on the seismic wave frequency spectrum by utilizing the following formula shown in the specification, so as to acquire the seismic wave frequency spectrum U (tau, omega) after inverse Q filtering, wherein gamma is equal to 1 / (phiQ); omega h is the highest frequency within the seismic bandwidth; sigma 2 is a stable factor; i is an imaginary part unit; tau' in exponential terms is an integration variable, and the value range of the integration variable is within an integrating range from omicron to tau; and the quality factor Q is related to the integration variable tau'. The high-frequency component is not pressed in the stabilizing treatment process in the inverse Q filtering method provided by the invention, so the inverse Q filtering method has the advantages that the resolution ratio is increased but the high-frequency noise is not increased, and the signal to noise ratio (SNR) is increased.

The invention discloses a bearing combined failure diagnosis method based on improved impact dictionary matching pursuit. According to the method, a bearing vibration signal is iterated and decomposed successively to be a linear combination based on an i-item atom of an improved impact model dictionary; an improved model is established according to the type of an analyzed bearing, and can very accurately indicate out an impact signal of a fault bearing in the running process; a position parameter u in the improved model is used as a preferentially changing parameter according to the cycle characteristic of a bearing failure signal, an atom base with relatively low redundancy is established by using a method of gradually changing the parameters, and the search speed of an optimal atom is greatly improved; a analyzed signal is cut in the signal decomposing process, the optimal atom is searched from atoms in the atom base, and a corresponding impact signal component can be obtained by impact component reconfiguration; and then the failure characteristic frequency of a bearing is obtained by time-frequency transform, and thus the failure diagnosis of the bearing is achieved.

The invention relates to a power transformer load tap changer switching contact slap fault diagnosis method and device. The diagnosis method includes that: firstly, real-time monitoring is carried outon tap changer switching action process, in the monitored signal, the latter half of time domain signal major energy is selected, so as to avoid the energy maximum point in time domain signal; secondly, time-frequency transformation analysis is carried out; thirdly, spectral line of the signal, which is higher in amplitude at the range of 800-1000Hz in frequency domain, is determined through analysis; fourthly, comparing is carried out, if the spectral line is obviously amplified compared with the normal condition of the tap changer, the tap changer is in the switching contact slap fault. Theinvention has the advantages that: the found fault characteristic value can accurately reflect tap changer switching contact slap fault, characteristic is obvious, repeatability is good, and the method has universality on tap changers of the same type; the load tap changer contact slap fault diagnosis technology is simple in engineering realization, and a sensor is convenient in installation; anda microcomputer is adopted for detection, processing and diagnosis, performance is reliable, realization is convenient, and diagnosis conclusion is visual.

The embodiment of the invention provides a down-mixing signal generating method which comprises the steps of: carrying out time and frequency conversion on a received left channel signal and a received right channel signal to obtain a frequency domain signal, dividing the frequency domain signal into a plurality of frequency bands; computing a channel energy ratio and a channel phase difference of each frequency band; computing a phase difference of a down-mixing signal and a first channel signal in each frequency band according to the channel energy ratio and the channel phase difference; and computing a frequency domain down-mixing signal according to the phase difference of the left channel signal, the right channel signal, the down-mixing channel and the first channel signal in each frequency band. According to the method, the quality of coding and decoding dimensional sound is effectively improved.

The invention discloses an indoor human body posture recognition method based on weighted combined distance time frequency transformation, and the method comprises the steps: firstly carrying out the MTI filtering of indoor human body posture data recorded by a UWB (ultra wide band) radar, extracting the data comprising the body posture, and selecting an effective distance gate; secondly obtaining a combined distance time frequency distribution map through the weighted combined distance time frequency transformation, and extracting the feature parameters of a body envelope in the map; thirdly determining the mapping relation between different postures and the feature parameters through a machine learning algorithm; finally determining the class of body posture according to the obtained mapping relation. The method is effective and feasible, is reliable in performance, and can accurately recognize different indoor body postures.

A receiver apparatus includes a channel estimator (b106) to perform channel estimation. The channel estimator includes an interim channel impulse estimating unit (b106-1) that calculates an interim channel impulse response estimation value, a path extracting unit (b106-3) that extracts a path for use in calculation of a channel impulse response estimation value from among paths forming the interim channel impulse response estimation values, a channel impulse response estimating unit (b106-4) that calculates the channel impulse response estimation value using the path extracted by the path extracting unit, and a frequency response estimating unit (b106-5) that converts the channel impulse response estimation value into a frequency response estimation value by time-frequency transforming the channel impulse response estimation value. In this way, a receiver apparatus capable of performing accurate channel estimation results.

The invention relates to a method for identifying a time-varying structure modal frequency based on a time frequency distribution map. The method comprises the following steps of: 1, acquiring structural dynamic response signals of an identified structure and setting sampling time and sampling frequency; 2, performing time frequency transformation on each response signal to obtain a time frequency distribution coefficient and drawing the time frequency distribution map; 3, writing the time frequency distribution coefficient into a corresponding energy distribution form and rearranging the coefficient as a column vector; 4, determining a time frequency distribution region corresponding to the response containing each-order time-varying modal frequency for identification according to the time frequency distribution map of each response; 5, extracting parts with the highest energy time frequency distribution corresponding to the each-order time-varying modal frequency from the time frequency distribution map by using proper time frequency window functions respectively; 6, estimating the each-order time-varying modal frequency by using a weighting nonlinear least square method; and 7, performing error analysis on the identification result. The method has the advantages of clear physical significance, simple and convenient use, high applicability and high anti-interference capability.

The acoustic impedance of the respiratory system can be inferred from oscillations that are generated in an airway of a subject. The impedance describes the frequency-dependent relation between the resulting oscillations in flow and pressure. When the impedance varies from inspiration to expiration, it has to be estimated with a high time resolution. A method is provided that reliably estimates the impedance in time intervals that are short enough for physiological purposes. A simple version of the uncertainty principle has been derived for discrete time and frequency. A discrete time-frequency transform has been developed that gives an optimal time-frequency resolution according to this principle. The transform is orthonormal, which permits an analysis of variance in the discrete time-frequency domain. The impedance follows from bivariate least-squares analysis in the time-frequency domain, under the assumption that noise is present in both flow and pressure.

The invention relates to a weak signal detection device based on time frequency conversion, which comprises a signal channel unit, an analog-digital conversion unit and a processor. A modulated signal to be tested is inputted into the signal channel unit for amplification and filtering by the signal channel unit; the analog-digital conversion unit outputs a quantized signal after sampling the analog signal outputted by the signal channel unit; the processor receives the quantized signal outputted by the analog-digital conversion unit and converts the quantized signal into a frequency domain signal for power spectral analysis, thereby obtaining the measuring results. The small signal detection device adopts the processor which converts the time domain signal into the frequency domain signal to detect the weak signal in the frequency domain and obtain the measuring results without a device producing a specific reference signal, thereby bringing about simple hardware structure and low cost.

The embodiment of the invention discloses a method, a device and equipment for evaluating speech quality, so as to lessen problems of high complexity and serious resource consumption of the existing signal domain evaluation scheme. The method of the embodiment of the invention comprises steps: the time domain envelope for speech signals is acquired; time-frequency transform is carried out on the time domain envelope to acquire an envelope spectrum; feature extraction is carried out on the envelope spectrum to acquire feature parameters; and according to the feature parameters, communication speech quality evaluation is carried out to acquire first speech quality parameters of the speech signals, and according to the first speech quality parameters and second speech quality parameters, comprehensive analysis is carried out to obtain the quality evaluation parameters of the section of inputted speech signals. The embodiment of the invention evaluates the quality of the speech signals in a condition of simulating auditory perception not based on a high-complexity cochlear filter, the calculation complexity is reduced, and the resource consumption is reduced.

The invention discloses a recording terminal clustering method based on voice time-frequency transform features and integral linear programming. The method comprises the following steps: a, carrying out preprocessing on read recording; b, extracting GTCC features from each frame of voices; c, constructing a depth self-coding network with a bottleneck layer, and extracting bottleneck features; d, constructing gaussian supervector features; e, carrying out spectrogram feature extraction on the voice obtained after preprocessing in step a; f, splicing the spectrogram features in step e to the gaussian supervector in step d, thus obtaining voice time-frequency transform features describing the features of the recording terminal finally; and g, carrying out clustering on the voice time-frequency transform features of all the recording samples by adopting the integral linear programming algorithm, wherein the clustering result is taken as the final classification basis. The feature extraction and clustering of the method provided by the invention are unsupervised, and compared with the existing supervising method, the method provided by the invention has better universality.

The invention belongs to the field of electronic reconnaissance identification, in particular to a radar emitter signal modulation identification method combined with multi-dimensional feature migration fusion, comprising the following steps of generating nine kinds of radar signals to form a radar signal set; transforming the radar signal into time-frequency image by time-frequency transform; transforming the time-frequency image so as to meet the input requirements of the pre-trained large-scale network; sending the pre-processed time-frequency image to LeNet 5 network for feature extraction, and using the feature extraction module from input layer to form C5 convolution layer to output the feature extraction module; selecting a dimensionality reduction mode for the data obtained from the extracting feature step and processing the dimensionality reduction mode. The invention adopts the method of time-frequency analysis, maps the one-dimensional time-domain signal to the two-dimensional time-frequency domain, analyzes and processes the radar signal in the time-frequency domain, and has better effect for the non-stationary radar signal. The self-training network adopted by the invention has simple structure, and can improve the reliability of the system under the condition of low signal-to-noise ratio.