78 results about "Sound source identification" patented technology

The invention relates to the technical field of intelligent robots, in particular to an intelligent human-shaped accompanying robot and a voice interaction system thereof, and the system comprises a voice input collection and preprocessing module which collects a composite voice signal and a voice triggering instruction in a surrounding environment in real time through a distributed microphone array; the multi-sound-source voice separation module is used for combining and independently processing the potential sound source signals and separating the potential sound source signals into a plurality of independent sound source signals; the target user sound source identification and extraction module is used for predicting the change of a target sound source in continuous time frames by utilizing a dynamic Bayesian network and extracting a target user sound source signal; the semantic analysis and output module is used for generating a natural and smooth voice response through a voice synthesis technology; according to the method, the problem that the target user sound source is difficult to extract in a noisy environment, multi-sound-source overlapping and voice switching scenes is solved, and the robustness and the adaptive capacity of the system in a complex voice environment are greatly improved.

The invention belongs to the technical field of sound source recognition, and particularly relates to a gas pipeline leakage sound source recognition method based on expansion delay summation, which comprises the following steps of: performing leakage sound source recognition on a scanning area by adopting a planar spiral microphone array to obtain a receiving signal; constructing a noise matrix Q according to the cross-spectrum matrix of the received signal p; based on the noise matrix Q, iteratively solving the position of each leakage sound source and the intensity of the leakage source by adopting an expansion delay summation method; according to the method, the base mismatching problem can be effectively relieved.

The invention discloses a bridge expansion joint acoustic monitoring method based on a beam forming microphone array, and the method comprises the steps: synchronously collecting acoustic signals through multiple channels, and achieving the high-precision signal alignment through clock offset compensation; the noise analysis module is started by combining a pre-triggering signal of a vehicle passing through the detection unit, the optimal microphone distance is dynamically determined according to sound field coherence and a signal-to-noise ratio maximization model, and a stepping motor drives a movable microphone to achieve real-time adjustment. On the basis, a spatial filtering weight is generated by adopting a self-adaptive beam forming algorithm, and directional enhancement and interference suppression of the expansion joint sound source are realized. Through collaborative optimization of a hardware structure and a signal processing method, the sensitivity and accuracy of acoustic monitoring of the bridge expansion joint are effectively improved, the abnormal sound source recognition capability can be remarkably enhanced, the environmental noise interference is reduced, and the method is suitable for intelligent monitoring and diagnosis of the long-term service state of the bridge expansion joint.

The invention provides a marine acoustic environment monitoring system based on a noise map. The marine acoustic environment monitoring system comprises a buoy network module, a data fusion processing module, a noise map generation module, a noise source identification module and an edge response module, the buoy network module serves as a sensing tail end, multi-source information is gathered to the data fusion processing module through the first data bus for standardization processing, and then the noise map generation module is driven through the second data bus to construct a three-dimensional sound field model. The output of the model is transmitted to a noise source identification module through a third data bus for intelligent analysis and traceability; the alarm decision of the noise source identification module is output to the edge response module, and the edge response module dynamically regulates and controls the communication and acquisition strategy of each buoy node through a distributed control circuit to form a complete closed loop. According to the invention, high-precision and automatic monitoring of the marine acoustic environment is realized, the problems of limited monitoring range and single data dimension of the traditional means are solved, and the reliability and emergency capability of the system are improved.

The invention discloses a construction and recognition method for a sound source recognition network of vehicles in a highway tunnel, and the method comprises the steps: setting a distributed microphone array in the tunnel, and collecting a vehicle audio signal; a MobileNetV3 model based on the Mel frequency spectrum is constructed; the Mel spectrum feature extraction module is used for taking the Mel spectrum feature extracted by the Mel spectrum feature extraction module as input, taking a vehicle audio signal type as output, and training a MobileNetV3 model based on the Mel spectrum to obtain a sound recognition model; constructing a sound source localization model; according to the method, the MobileNetV3 model based on the Mel frequency spectrum is adopted for sound recognition, the sound source positioning model is constructed in a combined mode for sound source positioning, the MobileNetV3 model and the sound source positioning model work cooperatively, the feature description capability of abnormal sound of a vehicle is enhanced, the influence of tunnel echo and environment noise on recognition is effectively reduced, accidents in the tunnel are found in time and positioned accurately, and the method is suitable for popularization and application. The recognition and positioning accuracy of the vehicle sound source in the complex tunnel environment is improved, and the technical problem that in the prior art, the recognition precision of the accident sound in the tunnel is not high is solved.

An active noise reduction system for an office environment relates to the technical field of noise reduction and comprises a sound source identification device, a microphone array, a noise reduction device, a loudspeaker array, a sound source detection module, a sound source positioning module, an environment evaluation module, a noise reduction control module, a central processing module and a user interface, according to the invention, the sound source identification and noise reduction equipment with the microphone array and the loudspeaker array is configured, and sound source detection, positioning, environment evaluation and noise reduction control modules are combined, so that real-time monitoring, accurate positioning, intelligent noise reduction and dynamic adjustment of the sound source are realized. The system can automatically adjust a noise reduction strategy according to station distribution and a sound source environment to ensure that an office environment is quiet and comfortable.

The invention is applicable to the technical field of grouting of mining engineering and geotechnical engineering, and provides a grouting diffusion space real-time sensing and tracking method based on slurry dynamic sound source localization, which comprises the following steps: active enhancement and characteristic modulation of a slurry dynamic sound source; building a stereo wave sensing network and acquiring high-fidelity sound wave data; carrying out AI-driven slurry sound source identification and three-dimensional space positioning; and the grouting diffusion process is subjected to three-dimensional real-time visualization and intelligent early warning. The invention creatively provides a comprehensive solution integrating sound source active enhancement modulation, stereophonic sensing network, artificial intelligence deep noise reduction and three-dimensional space inversion positioning, and the core of the comprehensive solution lies in that slurry flow sound signals which are originally weak and difficult to use are actively enhanced and are endowed with features; a traceable slurry power sound source signal with a high signal-to-noise ratio is constructed, and three-dimensional space real-time sensing and visual tracking of the slurry diffusion frontal surface position and the overall diffusion range are realized in combination with an advanced signal processing and positioning algorithm.

The present application relates to the technical field of noise diagnosis, in particular to a method and device for identifying abnormal sound source applied to a vehicle transmission system, the method comprising: acquiring abnormal sound noise data, and confirming an abnormal sound generation period according to the abnormal sound noise data; acquiring transmission system vibration data, and performing time-frequency domain analysis on the transmission system vibration data and the abnormal sound noise data according to the abnormal sound generation period to obtain abnormal sound time domain data; the transmission system vibration data is obtained by vibration detection of the vehicle transmission system at a previous time; performing cross-correlation identification of the abnormal sound source according to the abnormal sound time domain data to obtain an abnormal sound source identification result; the abnormal sound source identification result is used to represent that the abnormal sound source is a gearbox and / or a drive axle. By considering the state of strong coupling connection between the gearbox and the drive axle in the vehicle transmission system, cross-correlation analysis is performed on the abnormal sound time domain data to identify the abnormal sound source, thereby improving the accuracy of identifying the abnormal sound source of the vehicle identification system.

This invention discloses an ultra-high resolution compressed spherical beamforming sound source identification method, comprising the following steps: 1) constructing a two-dimensional off-grid compressed spherical beamforming model in the spherical harmonic domain based on Taylor expansion; 2) providing prior assumptions for the unknown variables in the two-dimensional off-grid compressed spherical beamforming model in the spherical harmonic domain, and calculating the joint probability density function of the unknown variables; 3) solving for the unknown variables in the two-dimensional off-grid compressed spherical beamforming model in the spherical harmonic domain according to the joint probability density function of the unknown variables, and estimating the sound source DOA and sound source intensity. This invention overcomes the basis mismatch problem, accurately estimates the off-grid sound source DOA and quantizes the sound source intensity, and achieves ultra-high resolution.

The invention discloses a sound field calculation method fusing ridge estimation-truncated singular value method, and the method comprises the steps: 1) constructing an acoustic inverse problem solving model based on an inversion formula of sound source intensity Q; 2) inputting the measured sound pressure column vector PH into an acoustic inverse problem solving model, and solving the acoustic inverse problem solving model by using an RE-TSVD algorithm to obtain sound source intensity; and 3) predicting the sound pressure of any field point according to the sound source intensity and the transfer matrix from the equivalent source point to the target field point. According to the method, side lobe interference can be effectively suppressed, main lobe information of a larger sound field is reserved, and the method has better sound source recognition and positioning performance.

The present invention provides an improved function beamforming method and related devices, which are related to sound source identification based on a microphone array. By establishing an array signal receiving model; calculating the FB output according to the array signal receiving model; calculating the PSF matrix of power-raising space transfer invariance according to the array signal receiving model; determining the spatial convolution information according to the FB output and the PSF matrix; transforming the spatial convolution information into a wave number domain product through FFT operation under periodic boundary conditions; constructing a difference function according to the wave number domain product and the FB output; calculating the difference function through a fast iterative shrinkage threshold method, solving the wave number leakage problem caused by the zero-filled boundary, improving the calculation accuracy, and further improving the imaging performance; sharpening the main lobe of the point spread function through exponential operation, and expanding the applicability of the assumption of spatial transfer invariance of the point spread function. Improve the imaging spatial resolution and dynamic range, and expand the effective imaging area. A technical effect of avoiding performance loss in the process of sound source identification is achieved.

The present application belongs to the field of sound field reconstruction and sound source identification technology, and discloses an equivalent source sound field positioning method and system based on sparse sound intensity measurement, which acquires the complex sound pressure and complex particle velocity at the measurement point on the holographic surface through a vector sensor array, and calculates the measured sound intensity vector; constructs a sound intensity transfer matrix based on the equivalent source method, establishes a linear measurement equation between the measured sound intensity vector and the equivalent source intensity vector; by using the spatial sparsity of the equivalent source intensity, the solving of the equivalent source intensity vector is converted into an L1 norm sparse optimization problem, and a convex optimization algorithm is used to solve it to obtain the optimal equivalent source intensity vector; based on the optimal equivalent source intensity vector, the acoustic quantity at any point in the target sound field is reconstructed by combining the free field Green function. The present application uses the active component of the sound intensity vector to suppress non-related noise, improves the ill-conditioned inversion by combining the compressive sensing theory, and can realize high robustness and high precision sound source positioning and sound field reconstruction without reference microphones.

The invention discloses a cable external damage early warning method and device based on sound source time-frequency domain feature identification, and relates to the field of cable external damage prevention. The method comprises the following steps: acquiring construction signal data; the construction signal data is sound signal data collected in real time around the marking pile; processing the construction signal data based on an amplitude section to obtain time domain feature data; determining frequency domain data; the frequency domain data is obtained by performing fast Fourier transform on the construction signal data; performing weighting processing according to the time domain feature data and the frequency domain data to obtain processing data; carrying out numerical comparison on the processing data and the sample data under the condition of setting the total number of collection and identification times, and carrying out sound source identification according to an obtained comparison result to obtain an identification result; the sample data is obtained by performing time domain feature calculation and weighting processing on the destroyed source signal data; and determining early warning information according to the identification result. The invention aims to realize the early warning of the external damage condition of the cable.

A method of presenting audio comprises: identifying a first ear listener position and a second ear listener position in a mixed reality environment; identifying a first virtual sound source in the mixed reality environment; identifying a first object in the mixed reality environment; determining a first audio signal in the mixed reality environment, wherein the first audio signal originates at the first virtual sound source and intersects the first ear listener position; determining a second audio signal in the mixed reality environment, wherein the second audio signal originates at the first virtual sound source, intersects the first object, and intersects the second ear listener position; determining a third audio signal based on the second audio signal and the first object; presenting, to a first ear of a user, the first audio signal; and presenting, to a second ear of the user, the third audio signal.

This invention discloses a construction area vibration zoning control method based on multi-source sound identification, relating to the field of engineering construction control. The method includes: acquiring the coordinate positions of the target construction area and the railway track, and determining the upper limit of vibration intensity in the target construction area; determining the target set of construction equipment, and based on the construction movement line, determining the fixed positions of the foundation construction equipment set within the target construction area to form a foundation equipment layout scheme, and generating a foundation construction vibration distribution through multi-source sound identification analysis; performing vibration zoning on the foundation construction vibration distribution, and based on the vibration zoning results and the upper limit of vibration intensity, optimizing the equipment layout of the movable construction equipment set based on the foundation equipment layout scheme to generate a target equipment layout scheme. This solves the technical problem in existing technologies where regional vibration load imbalance prevents the optimal allocation of vibration energy in the construction area, leading to both low construction efficiency and potential safety risks.

This invention relates to a method and system for locating coaxial counter-rotating sound sources in rotorcraft, belonging to the technical field of sound source identification methods. It includes: processing the sound pressure signal of the moving sound source using continuous wavelet transform to obtain time-frequency domain variation results; performing Doppler effect correction based on experimental conditions; estimating the spatial location and relative intensity of the noise source using a cross-spectral matrix and weight vector; and performing phase-locked analysis of the rotor noise source intensity using a phase-locked averaging method. This invention differs from traditional time-domain delay superposition algorithms and frequency-domain cross-correlation spectrum methods. It can image unsteady sound sources using time-frequency domain characteristics, solving the acoustic imaging problem of counter-rotating propeller noise and possessing potential applications in the rotorcraft industry.

The invention is suitable for the field of intelligent sound box control, and provides a broadcast control method and system for an intelligent sound box, and the system comprises a sound collection and direction judgment module, a sound feature extraction and separation module, a voice recognition and content analysis module, a region division and priority judgment module, and a broadcast content generation and region binding module. According to the scheme, voice input in multiple directions can be received at the same time, and the direction of each sound source can be accurately recognized. The capability remarkably improves the operation efficiency of the sound box in a complex environment, and can meet the requirement that multiple persons send out instructions at the same time. Through analysis of the sound intensity and the time difference, the sound box can accurately identify the position of each sound source, and the space is divided into different areas. According to the dynamic space management mode, the sound box can be intelligently adapted according to the actual environment, and the accuracy of sound source identification and the correlation of response are improved.

This invention discloses an FPGA implementation method for visual image processing of sound source identification, comprising: 1. Normalizing the sound intensity values ​​obtained from sound source identification and converting them into grayscale values; 2. Obtaining corresponding color pixel values ​​by performing three independent mappings on the grayscale values; 3. Upsampling the low-resolution color acoustic cloud map using an interpolation algorithm to improve the image resolution and form a visual acoustic cloud map that represents sound intensity in color. This invention enables the visual image processing of sound source identification to be implemented on an FPGA in the form of digital circuits, thereby improving the real-time performance of the entire sound source identification process, reducing the hardware complexity of the sound source identification platform, and saving time and hardware costs.

The invention discloses a rotating sound source recognition and positioning method based on a deep convolutional attention neural network, and belongs to the field of rotating sound source recognition, and the method comprises the steps: firstly carrying out the motion compensation and modal expansion of a sound pressure cross-spectrum matrix through a modal decomposition rotating beam forming algorithm, and generating a sound source distribution diagram in a rotating scene; then the sound source distribution map is input into a deep convolution attention neural network model, the model adopts an encoder-decoder structure, a channel and a space attention mechanism are fused, key sound source features are dynamically focused, noise interference is suppressed, and high-precision reconstruction of the sound source distribution map is achieved; and finally, carrying out local maximum value detection on a prediction map output by the network, and extracting a sound source space position. According to the method, the Doppler effect and phase distortion problems in rotating sound source positioning are effectively overcome, and the positioning precision, the anti-noise performance and the calculation efficiency in a complex rotating field are remarkably improved.

The invention belongs to the technical field of acoustic measurement, and discloses a free space sound field measurement method and system, and the method comprises the steps: constructing a free light path sound field measurement system based on a Michelson interferometer principle; acquiring a reference image, introducing a sound field to be measured, and controlling a loudspeaker or a laser beam to perform translation and rotary scanning through a computer to complete interference pattern data acquisition; preprocessing the acquired interference pattern, identifying the edge of the interference fringes by using an edge detection technology, and carrying out light phase variation calculation; and performing sound field reconstruction by adopting a Radon inverse transformation algorithm, converting sound field integral information carried by a laser beam into a sound pressure value of each point in a real space, and reconstructing a free field sound pressure distribution diagram. The method has the advantages of no damage and low cost, the measurement precision of the sound field and the integrity and reliability of reconstruction are remarkably improved, the nano-scale optical path change caused by the sound field can be effectively captured, and the high precision of sound field measurement and the accuracy of sound source recognition are ensured.

The invention discloses an orthogonal matching pursuit sound source identification and positioning method based on a feedback mechanism, and the method comprises the steps: setting a measurement surface and a focusing surface according to a set distance interval, carrying out the grid division of the measurement surface and the focusing surface, and obtaining a microphone position, a focusing point and a Green function transfer matrix; obtaining the relation between the sound pressure vector measured by the microphone and the source intensity of the sound source at the focus point by using the Green function transfer matrix; and the sound source intensity of the focusing surface grid points is calculated through an orthogonal matching pursuit algorithm based on source intensity prior guidance, and the sound source is accurately identified and positioned by using the obtained sound source intensity data. According to the method, the atoms are preliminarily selected through the orthogonal matching pursuit algorithm guided by the source intensity prior information, the preliminarily selected atoms are fed back to the transfer matrix, the atoms closest to the measured sound pressure vector are obtained to serve as the finally selected atoms, the reconstruction performance of a traditional algorithm at low and medium frequencies is improved, and the method is suitable for being applied to the field of sound pressure measurement. And meanwhile, the sound source identification and positioning capability under the strong correlation between atoms of the measurement matrix is improved.

The invention discloses a scenarized audio recognition method and device, electronic equipment and a storage medium, and relates to the technical field of artificial intelligence sound signal processing. The method comprises the following steps: inputting a collected original audio waveform into a shared feature extraction network for feature extraction processing to obtain a first audio feature map; performing prior information processing based on the first audio feature map and the external input vector to obtain dynamic sound source prior information; the dynamic sound source prior information comprises an internal activeness label vector and an external sound source activeness vector; performing parameter generation processing according to the internal activeness label vector and the external sound source activeness vector to obtain a feature modulation parameter; performing affine transformation processing on the first audio feature map according to the feature modulation parameter to obtain a modulated second audio feature map; and inputting the second audio feature map into a sound source classifier for identification processing to obtain a sound source identification result. The environment dynamic adaptability and accuracy of scenarized audio recognition can be improved.

This application discloses an integrated monitoring boundary marker system based on multimodal sensing and redundant transmission, relating to the field of nature reserve monitoring technology. The system includes multiple boundary markers, each containing sensing and imaging equipment, a dual-power management unit, a sound source identification unit, a multi-source data fusion processing unit, and a multi-mode communication unit electrically connected to the system. This system can effectively monitor and accurately transmit monitoring data for protected areas with an average area of ​​82,000 to 83,000 hectares when only 58 to 59 boundary markers are used. It offers high monitoring reliability and reduces data mistransmission.

This invention relates to a sound source identification method based on array expansion and SE (Search Engine Attention) mobile inverse bottleneck convolutional neural network, comprising: acquiring an 18-array sound source distribution map calculated using the MUSIC algorithm based on the sound pressure cross-spectrum matrix; inputting the 18-array sound source distribution map into an EAG-U-Net data conversion model to convert the 18-array sound source distribution map into a 64-array sound source distribution map, achieving the effect of expanding the microphone array. This model introduces the EAG mechanism to optimize model feature selection in the spatial and channel dimensions, improving the model's representational ability and accuracy in the data conversion process, and generating an acoustic imaging map after data conversion; inputting the 64-array sound source distribution map into a mobile inverse bottleneck convolutional neural network model based on the SE attention mechanism for feature extraction and image reconstruction to obtain a predicted sound source distribution map; and performing local maxima detection using the predicted sound source distribution map to obtain sound source localization and intensity.

Soundproof chamber construction equipment (100, 200, 300), comprising: an audio procurement unit (101, 201) that procures audio data that includes audio from a variety of sound sources; a sound source identification unit (102, 202) which, based on the audio data, determines a plurality of sound source positions as positions of the plurality of sound sources; an audio extraction unit (103) that generates a multitude of pieces of extraction audio data by extracting audio represented by the audio data with respect to each sound source and generating the extraction audio data representing the extracted audio; a format conversion unit (104) that generates a multitude of stereophonic tones corresponding to the multitude of sound sources by converting a format of the multitude of pieces of extraction audio data into a stereophonic audio format; a position procurement unit (105) that procures a listening position as a position at which audio is listened to; a motion processing unit (106) that calculates an angle and distance between the listening position and each of the multitude of sound source positions; an angle-distance adjustment unit (107) that adjusts each of the plurality of stereophonic tones using the angle and distance according to each of the plurality of sound source positions, thereby producing a plurality of set stereophonic tones as a plurality of stereophonic tones at the listening position; and a superposition unit (108, 308) that superimposes the multitude of set stereophonic tones.

The method of operation of a wearable device according to the present invention comprises: a step in which a microphone array module acquires noise magnitude information and direction information; a step in which a wireless communication module transmits the noise magnitude information and direction information to an external cloud server via a user's smart device under the control of a processor; a step in which the wireless communication module receives noise source identification information and risk level information corresponding thereto from the external cloud server via the smart device under the control of a processor; a step in which the processor generates a control signal to provide at least one stimulus among vibration stimulation, electric stimulation, and visual stimulation by a light-emitting element to a user using at least one of the noise magnitude information, the noise direction information, the noise source identification information, and the risk level information corresponding thereto; and a step in which at least one of a vibration element, an electric stimulation array, and a light-emitting element operates to generate at least one stimulus according to the control signal.

The embodiment of the invention provides an equipment state detection method and device, equipment and a storage medium, and relates to the field of artificial intelligence. The equipment state detection method comprises the following steps: acquiring displacement information of equipment through a gyroscope module, and determining a motion state of the equipment according to the displacement information; if the motion state of the equipment is a static state, acquiring environment sound through a recording module, and identifying the environment sound to obtain a sound source identification result; determining an equipment use condition of the equipment according to the sound source identification result, and determining a working state of the equipment according to the equipment use condition of the equipment; acquiring position information of the equipment or environment information of the equipment according to the working state of the equipment; according to the position information of the equipment or the environment information of the equipment, the equipment state of the equipment is determined, and the equipment state comprises an idle state and a lost state. The invention aims to solve the problems of low equipment utilization rate and high equipment loss risk in a specific scene.

The embodiment of the invention provides an acoustic horn immersion type sound field rendering method and system, and relates to the technical field of sound field rendering, and the method comprises the steps: obtaining a user position and a head posture in a virtual space; acquiring the position and intensity of a virtual space sound source; identifying a focus sound source currently concerned by the user according to the user position, the head posture, the sound source position and the intensity; identifying sound sources except the focus sound source as environment sound sources; performing sound field rendering on the environmental sound source to obtain a shared environmental sound field signal, the computing resource of the environmental sound source being lower than the computing resource of the focus sound source; performing convolution operation on the original audio signal of the focus sound source and the head acoustic transmission function to obtain a personalized high-resolution signal; and superposing the shared environment sound field signal and the personalized high-resolution signal to generate a binaural signal, and playing the binaural signal through an acoustic horn array. According to the method, the immersion and the actual effect of virtual reality training can be improved.

The application discloses a kind of orthogonal matching pursuit sound source identification and positioning method based on feedback mechanism, measurement surface and focusing surface are set according to set distance interval, measurement surface and focusing surface are grid division, obtain microphone position and focusing point and Green function transfer matrix;The relationship between microphone measurement sound pressure vector and focusing point sound source intensity is obtained using Green function transfer matrix;The sound source intensity of focusing surface grid point is calculated by orthogonal matching pursuit algorithm based on source intensity priori guidance, and the obtained sound source intensity data is used for accurate identification and positioning of sound source.The application preliminarily selects atom by orthogonal matching pursuit algorithm based on source intensity prior information guidance, then the atom preliminarily selected is fed back to transfer matrix, and the atom closest to measurement sound pressure vector is obtained as the final selected atom, which improves the reconstruction performance of traditional algorithm at low frequency, and improves the sound source identification and positioning ability under the strong correlation between measurement matrix atoms.

The invention discloses a noise source identification test device for an axial flow compressor of a turboshaft engine, belongs to the technical field of aerodynamic noise analysis, and aims to solve the problems of poor real-time performance and low accuracy in noise identification of the axial flow compressor of the turboshaft engine. The axial flow compressor testing system is composed of a single-stage axial flow compressor model and a driving motor, the driving motor is connected with a compressor rotating shaft through a high-precision vibration reduction coupling, and a motor shell is covered with sound absorption cotton; the noise measurement system comprises a semicircular microphone array and a temperature acquisition unit, and a temperature sensor is mounted at a key position of a gas compressor; the noise source identification system comprises a high-speed acquisition module and a hybrid intelligent identification module; the acquisition module performs time domain filtering and spectral analysis preprocessing on original sound signals collected by a microphone, and the hybrid model realizes noise source real-time classification through multi-working-condition noise sample end-to-end training.