11 results about "Spectral flatness" patented technology

The invention provides a sound wave optical cable positioning method and system, and relates to the technical field of deep learning, and the method comprises the steps: uniformly arranging vibration sensors on a sensing optical cable, demodulating a back scattering light signal to obtain a differential phase signal, and constructing a self-correlation loss function. Preprocessing sound source data, extracting an audio frequency semantic feature vector, and calculating a dot product of a frequency spectrum centroid and frequency spectrum flatness to obtain a frequency spectrum loss function; and a phase change loss function is obtained by combining the ratio and the refractive index of the strained optical fiber and the non-strained optical fiber. And weighting the three loss functions to form a fusion loss function. And constructing a sound wave positioning model, and training the model to be optimal based on the fusion loss function. And finally, segmenting and enhancing audio data, and inputting the audio data into the trained model to obtain a positioning result. According to the method, the positioning model is trained by fusing multiple different types of loss functions, so that the positioning precision of sound waves in the optical cable is improved.

The application relates to the field of audio transmission technology and discloses an FD-CAN bus audio composite data stream transmission method and system. Three types of real-time multi-modal parameters, including audio content features, bus physical features and service quality constraints, are acquired, an audio dynamic complexity coefficient is calculated according to audio instantaneous energy and spectral flatness, a bus transmission quality coefficient is calculated according to bus instantaneous load rate and signal-to-noise ratio, the maximum transmission delay allowed is obtained according to service quality and signal-to-noise ratio, a dynamic scheduling priority coefficient is acquired in combination with the bus transmission quality coefficient, a time-frequency-space multi-modal coupling matrix is constructed based on the three types of coefficients, a principal coupling feature vector is extracted through principal component analysis and a dynamic transmission strategy index value is calculated, a frame format, a modulation mode and error correction coding parameters are adaptively selected according to the index value, and the audio is sent after being packaged into an FD-CAN frame. The application solves the problems of bandwidth waste and transmission jitter caused by traditional static scheduling through multi-dimensional parameter coupling and dynamic strategy adaptation.

An audio processing method includes: filtering an audio signal to obtain a low-frequency signal and a high-frequency signal; encoding the low-frequency signal to obtain a bitstream of the low-frequency signal; performing frequency domain transform on the low-frequency signal and the high-frequency signal respectively, to obtain a low-frequency spectrum and a high-frequency spectrum; performing spectral envelope extraction on the low-frequency spectrum and the high-frequency spectrum to obtain spectral envelope information, and performing spectral flatness extraction on the high-frequency spectrum to obtain spectral flatness information; and performing quantization encoding on the spectral flatness information and the spectral envelope information to obtain a bandwidth extension bitstream, and combining the bandwidth extension bitstream and the bitstream of low-frequency signal into an encoded bitstream.

This invention discloses a blind calibration and interference cancellation method for frequency-varying errors based on spectral flatness. The received signal is divided into frequency-domain sub-bands. Within each sub-band, spectral flatness is used as the optimization criterion, and frequency-varying amplitude and phase compensation parameters are introduced as optimization variables into a genetic algorithm for global optimization. Error compensation is applied to the auxiliary antenna signal based on the optimized parameters. Interference cancellation is then achieved within the compensated sub-bands. Finally, sub-band merging is performed to recover the target signal. This method requires no prior information and achieves adaptive calibration of frequency-varying errors through optimization, effectively improving interference suppression performance.

The application discloses a speech structure sensitive area constraint enhancement method and system for a coal mine scene, and the method comprises the following steps: obtaining an original amplitude spectrum and a high-order statistical quantity feature vector corresponding to a noisy speech signal, and performing splicing processing on the two to obtain a composite tensor; time dimension features and frequency dimension features are extracted based on the composite tensor, a preliminary enhanced amplitude spectrum is further obtained, and a joint feature tensor comprising a spectral centroid change rate, spectral flatness, an inter-frame correlation coefficient and a medium frequency energy proportion is constructed; a sensitive probability graph is constructed based on the joint feature tensor, which is isomorphic with the original amplitude spectrum; a suppression penalty term is constructed according to the sensitive probability graph, the preliminary enhanced amplitude spectrum and a pure speech signal amplitude spectrum; and a preset speech deep learning model is optimized and trained by using the suppression penalty term. The application can consider both the suppression of non-stationary transient noise and the preservation of the sensitive area of the speech structure, and solves the problem of the decline of speech quality and communication intelligibility under extreme working conditions.

The invention discloses an intelligent hearing aid noise reduction method combined with environmental noise evaluation, and relates to the technical field of hearing aid signal processing, comprising the following steps: collecting an audio signal and converting the audio signal to a frequency domain to obtain a frequency domain signal; extracting noise features from the frequency domain signal, calculating a noise time-varying index and spectrum flatness, and identifying an environmental noise type; estimating a noise power spectrum, determining a noise tracking parameter according to the environment noise type, and updating the noise power spectrum; calculating a signal-to-noise ratio parameter corresponding to the noise power spectrum, determining a noise reduction intensity factor according to the environmental noise type and the hearing characteristics of the user, and calculating a noise reduction gain based on the signal-to-noise ratio parameter and the noise reduction intensity factor; and performing noise reduction processing on the frequency domain signal by using the noise reduction gain, and converting the frequency domain signal to a time domain to output a processed audio signal. According to the invention, the noise tracking parameter is adaptively adjusted by identifying the environmental noise type, and the noise reduction intensity is determined in combination with the hearing characteristics of the user, so that the noise reduction effect and the user suitability of the hearing aid in a complex environment are improved.

The application provides a time-frequency cross signal representation and extraction method and device based on time-frequency analysis, and relates to the technical field of signal processing. The method comprises the following steps: obtaining a signal to be analyzed and performing modulation transformation based on discrete modulation parameters; performing block division on a time-frequency result, calculating the spectral flatness of each time-frequency block, selecting an optimal modulation parameter, and obtaining a spectral flatness matching modulation transformation result based on the optimal modulation parameter; extracting a time-frequency ridge based on the result and constructing an extraction operator; and performing time-frequency coefficient extraction by using the extraction operator to obtain a representation result. The application uses spectral flatness as an evaluation index, effectively avoids energy ambiguity caused by local optimization, constructs an extraction operator based on a ridge, solves the feature missing problem in a time-frequency cross region, realizes high-precision and high-energy-concentration signal representation and extraction, and is suitable for aero-engine rotor fault diagnosis.

This application discloses a signal quality improvement method and a waveform generator. The signal quality improvement method includes acquiring preset output signal parameters of the waveform generator; determining a current compensation coefficient corresponding to the preset output signal parameters; and performing flatness compensation on the digital baseband signal generated by the waveform generator according to the current compensation coefficient. The current compensation coefficient makes the frequency response characteristics of the signal output by the waveform generator after compensation exhibit a state opposite to the frequency response characteristics of the transmission link, thereby improving the spectral flatness of the signal output by the waveform generator after passing through the transmission link. Since the frequency response characteristics of the compensated digital baseband signal after passing through the transmission link are consistent with the frequency response characteristics of the digital baseband signal before compensation, the compensated digital baseband signal can accurately reflect the frequency response characteristics of the device under test when output to the device under test through the transmission link.

The invention discloses a method and a system for establishing and evaluating a sound quality objective data set of motor unsteady-state working condition noise, and the method comprises the steps: collecting a noise signal, calculating a variation coefficient and a frequency spectrum flatness change rate, eliminating an abnormal value through a quartile method, and determining a steady-state condition. And under the non-steady-state working condition, if the adjacent noise segments meet the steady-state condition, merging is carried out, and otherwise, segmentation is carried out. The method comprises the following steps: extracting loudness, roughness, sharpness, voice definition and A-weighted sound pressure level based on a psychological acoustic model, calculating a weighted average value to represent unsteady noise features, and constructing a feature matrix; in order to optimize sample grouping, kernel principal component analysis is adopted to reduce dimensions to three dimensions, k-means + + clustering is utilized, a greedy optimization strategy is combined, and only alpha samples closest to the centroid are reserved to optimize grouping. After grouping is completed, paired comparison is carried out in the groups, score inversion is carried out on the groups with the rear rankings, and the integrity and consistency of results are ensured.

The invention discloses a signal quality improving method and a waveform generator. The signal quality improving method comprises the following steps: acquiring a preset output signal parameter of the waveform generator; determining a current compensation coefficient corresponding to the preset output signal parameter; and according to the current compensation coefficient, flatness compensation is carried out on the digital baseband signal generated by the waveform generator, and the current compensation coefficient enables the frequency response characteristic of the signal output by the waveform generator after compensation to be in a state opposite to the frequency response characteristic of the transmission link. Therefore, the spectrum flatness of the signal output by the waveform generator after passing through the transmission link is improved. As the frequency response characteristic of the compensated digital baseband signal after passing through the transmission link is consistent with the frequency response characteristic of the digital baseband signal before compensation, when the compensated digital baseband signal is output to the tested equipment through the transmission link, the frequency response characteristic of the tested equipment can be correctly reflected.