126 results about "Quantization (signal processing)" patented technology

A video signal decoding device comprises a processor which: determines a particular quantizer for reconstructing a first quantized transform coefficient, the particular quantizer being one of a first quantizer and a second quantizer which are different from each other, the particular quantizer being determined on the basis of the state of the first quantized transform coefficient; reconstructs the first quantized transform coefficient on the basis of the particular quantizer to obtain a reconstructed transform coefficient; and updates the state of a second quantized transform coefficient that is reconstructed after the first quantized transform coefficient, wherein the first quantized transform coefficient and the second quantized transform coefficient are transform coefficients in the current block.

The invention discloses an audio signal processing method and device based on MOE, a medium and a product, and relates to the field of audio signal processing. According to the technical scheme, firstly, the audio signal processing device decomposes the audio signal into the sub-band signals in different frequency intervals, and the complexity parameter of each sub-band signal is calculated; on the basis of the sub-band signals and complexity parameters corresponding to the sub-band signals, the gating network in the hybrid expert model can intelligently distribute the optimal expert network combination for each sub-band signal, the speciality of each expert network is fully utilized, the processing efficiency is improved, the optimal expert network combination is screened by adopting a processing weight coefficient threshold, and the processing efficiency is improved. And redundant processing is avoided. And finally, the audio signal processing device obtains a structured audio data stream through quantization coding and data packaging. According to the dividing and treating method, the quality and efficiency of audio signal processing are improved, the processing strategy can be flexibly adjusted according to the signal characteristics, and the adaptability is high.

The invention relates to the technical field of intelligent signal processing, in particular to an unmanned aerial vehicle cluster intelligent beamforming method and system, and the method comprises the following steps: obtaining unmanned aerial vehicle node positions, antenna deviation and synchronization errors, analyzing the influence of quantization precision on beamforming, extracting low-precision waveform features, recognizing the amplitude-phase error compensation amount, and optimizing the anti-interference performance. Phase constraints are extracted and optimized, the consistency of the beam and the expected direction is compared, parameters are adjusted to obtain an adjustment instruction set, and a beam forming maintaining state label is output. According to the method, position information, pointing deviation and synchronization errors are analyzed through joint modeling, amplitude-phase feature extraction precision is improved, interference and echo separation capability is enhanced, control robustness is improved, discrete phase constraint and alternating direction optimization are introduced, control vector solution and convergence are accelerated, and direction errors are dynamically adjusted. A retention label is generated through convergence trend analysis, beam consistency and anti-interference performance are enhanced, and high-precision stable control is realized in a complex change scene.

The invention relates to the related field of underwater acoustic signal processing, in particular to an underwater acoustic target radiation noise classification method, which constructs an efficient and accurate underwater acoustic target radiation noise classification model through the technologies of lightweight model selection, feature fusion, transfer learning, hyper-parameter optimization, model pruning, quantification and the like. The model not only can realize real-time processing in an environment with limited resources, but also has good classification performance, low energy consumption and high flexibility, is suitable for various application occasions, and particularly has remarkable advantages in an underwater acoustic target classification and real-time monitoring system.

The invention relates to the technical field of image signal processing, and discloses an agent-free ISP hyper-parameter high-speed optimization method and system, and the method comprises the following steps: collecting RAW data through a sensor, carrying out the black level correction, and initializing the ISP register parameter mapping; a multi-dimensional state vector is constructed by combining the coded image and the parameters, and the master and slave agents generate discrete adjustment amounts in parallel; after a register is updated through nonlinear quantization, a composite reward is calculated based on target detection and image quality, and data is stored and network parameters are updated through a priority experience pool; the system comprises a sensor module, an ISP hardware processing unit, a master-slave agent decision module, a reward calculation module, a register mapping module and an experience playback module. According to the method, hardware-in-the-loop and deep reinforcement learning are combined, and the problems of proxy model distortion, response lag, high data dependence and parameter mutation are solved through master-slave agent parallel decision, a self-supervision reward mechanism and nonlinear parameter mapping.

The invention relates to the technical field of vital sign anomaly detection, in particular to a vital sign anomaly feature extraction method based on a sign signal, which comprises the following steps: firstly, dynamically adjusting the size of a time cabin through variational optimization, taking both transient and steady feature capture into account, and avoiding feature loss caused by a fixed window length; in combination with principal component analysis, Lyapunov exponential chaos quantization and a high-order differential operator, nonlinear features are decoupled from multiple dimensions of a time domain, a frequency domain and a phase domain, the feature interpretability is improved, further, through phase-speed mapping and attention weighted fusion, self-adaptive enhancement and decoupling of spatial-temporal features are achieved, noise interference is suppressed, and the spatial-temporal features are enhanced. The finally generated multi-dimensional feature cube integrates time resolution, frequency precision and phase differential information, the anomaly detection robustness of complex vital signs is remarkably improved, and the defects that a traditional method is single in dimension and insufficient in adaptability in dynamic signal processing are overcome.

The invention provides a single-bit SAR signal recovery method and device based on sparsity judgment, and relates to the technical field of signal processing, and the method comprises the steps: carrying out the range migration correction of a to-be-recovered single-bit quantization satellite-borne SAR echo signal, and obtaining a two-dimensional linear frequency modulation signal; performing space grid division on the target imaging area, and establishing an SAR imaging observation model in grid points obtained by division according to the two-dimensional linear frequency modulation signal; establishing a sparseness optimization model according to the SAR imaging observation model, and performing iterative updating on the sparseness optimization model to obtain an optimal estimated value of the scene sparseness; and recovering the corrected echo signal according to the optimal estimation value to obtain a reconstructed echo signal conforming to the quality standard. According to the scheme, the problem that target scene sparsity is difficult to obtain in the imaging processing process can be solved.

The invention relates to an analog-digital hybrid SRAM (Static Random Access Memory) memory audio noise reduction method and system, and the method comprises the following steps: carrying out the differential sampling quantization processing of an input audio signal, and obtaining a multi-bit-width digital audio code stream; inputting the multi-bit-width digital audio code stream into a preset SRAM (Static Random Access Memory) memory chip to carry out mixed mode write-in operation to obtain an analog-digital mixed memory matrix; performing frequency domain feature extraction on the analog-digital hybrid storage matrix to obtain frequency domain extraction features; performing noise identification and suppression on the audio signal based on the frequency domain extraction feature to obtain a time-varying noise suppression coefficient; and performing audio signal recovery processing on the time-varying noise suppression coefficient to obtain a denoised output audio signal, thereby solving the technical problem that power consumption and hardware cost are greatly increased due to the fact that a traditional digital signal processing method generally depends on a high-precision analog-to-digital converter and a high-performance processor.

The invention relates to the technical field of signal processing, and particularly provides a full-quantization spectrum characteristic statistical method, system and device and a storage medium, and the method comprises the steps: discretizing a current sampling signal frequency band into a continuous frequency point unit sequence according to a preset rule, and mapping a signal intensity value into an intensity level according to the preset rule; if the signal intensity exceeds the current noise base value, determining the frequency point range of the current sampling signal by extracting the center frequency point and the effective bandwidth of the current sampling signal, and recording the intensity level mapped for all frequency point units in the range once; if the signal intensity does not exceed the current noise base value, recording the intensity grade of mapping for one time for a frequency point unit in the frequency point unit sequence of the current sampling signal; and counting all involved frequency point units in the frequency spectrum recording process and the occurrence frequency of each intensity level recorded on each frequency point unit. According to the invention, the problem of virtual high background noise caused by a traditional peak value holding mode is solved.

Examples relate to the field of radio frequency (RF) signal processing such as classifying RF signals with low latency. The method involves receiving portions of an RF signal, transforming these portions into a time-resolved frequency representation using a continuous wavelet transform, and processing this representation with a recurrent neural network. The neural network modifies a neural network state incrementally to generate a classification output, which may include modulation classification, signal-to-noise ratio (SNR) classification, or jamming detection. The system achieves sub-millisecond inference latency through techniques such as model quantization and batch size optimization. Principal uses include real-time RF signal analysis and jamming detection, with applications in communication systems and environmental monitoring. The RF signal may be received from a quantum RF sensor based on Rydberg atoms, enabling broad frequency range detection.

The invention relates to the field of language signal processing, the field of financial science and technology and the field of medical health, and discloses a text-to-speech conversion method, device, equipment and medium, and the method comprises the steps: carrying out text word segmentation processing and part-of-speech tagging processing on a target text to obtain a preprocessed text, generating an output code according to the pre-processed text by using a preset double-autoregression architecture, generating a Mel spectrum according to the output code by using a decoder, performing quantization reconstruction processing on the Mel spectrum by using an encoding architecture to obtain a quantization tensor, calculating a tensor loss value of the quantization tensor and the Mel spectrum, and calculating the tensor loss value of the quantization tensor and the Mel spectrum; and optimizing parameters of the encoder architecture according to the tensor loss value based on a back propagation algorithm, generating a prompt code according to a pre-acquired to-be-processed text based on the optimized encoding architecture, generating a to-be-processed Mel spectrum in combination with the prompt code and the to-be-processed text, and generating a target audio according to the to-be-processed Mel spectrum by using a preset vocoder. And the efficiency and accuracy of converting the text into the audio are improved.

The invention discloses an AI-driven vehicle-mounted sound field real-time modeling and voice separation method, and relates to the technical field of voice signal processing. A main control unit comprising a time sequence synchronizer, a resource scheduler and a health monitor is constructed. 3D sound field modeling is carried out by adopting a lightweight STCN + bidirectional LSTM network, adaptive updating of the model is realized through EWC incremental learning, and a CNN-LSTM noise classification network and targeted suppression algorithms such as ANF / spectral subtraction are developed. The voice separation module adopts an improved Conv-TasNet architecture, 3D spatial constraint and a multi-task loss function are fused, and low delay is realized under INT8 quantization and pipeline processing. The system dynamically optimizes parameters through a real-time regulation and control unit, supports scene self-adaption, finally achieves a separation effect in a mixed noise scene, reduces the delay of the whole system, and effectively improves the definition and stability of vehicle-mounted voice interaction.

The invention discloses a controllable sparse migration attack method based on generator generation, and belongs to the crossing field of communication signal processing and artificial intelligence. The method aims at solving the problems that disturbance of existing countermeasure attacks in the field of electromagnetic signals is too concentrated, concealment is insufficient, and disturbance density is difficult to control. The technical scheme comprises the following steps: simultaneously outputting disturbance amplitude and disturbance probability by using a deep generation network; classifying key regions by calculating group frequency spectrum saliency positioning pairs and generating a group mask matrix; generating a binary position matrix through a mask quantization network in combination with the group mask and the disturbance probability; training the generative network by adopting a total loss function comprising adversarial loss and smooth loss; and finally, generating a sparse controllable adversarial sample through the trained network. According to the method, the adversarial sample which is high in concealment, high in mobility and controllable in disturbance proportion can be generated, and an effective means is provided for security assessment of the deep neural network.

The invention relates to an ultra-low power consumption end side AI audio signal processing method and system. The method comprises the steps of performing framing and logarithmic compression processing on an input audio signal to obtain target spectrum data, and decomposing the target spectrum data to obtain a modulation period signal and a non-modulation signal; performing critical frequency band analysis and variational Bayesian fixed-point operation on the non-modulated signal to obtain a quantized noise parameter; extracting a main period characteristic of the modulation period signal and a frequency band characteristic of the quantization noise parameter, and constructing a target discrimination characteristic according to the main period characteristic and the frequency band characteristic; and inputting the target discrimination feature into a radial basis support vector machine, and training by adopting a piecewise linear kernel function and a sequence minimum optimization algorithm of second-order Taylor expansion to obtain an audio processing model. According to the invention, accurate separation of modulation period signals and non-modulation signals is realized, and efficient utilization of system resources and accurate control of power consumption are realized.

The application provides a pulse density modulation interface implementation method and device of an analog-to-digital converter, electronic equipment and a storage medium. The method comprises the following steps: designing a circuit architecture of an analog-to-digital converter, and setting working variable parameters for components in the circuit architecture; determining a transfer function of an output signal and a transfer function of noise based on the working variable parameters; in the case that an oversampling rate (OSR) is a first set value, a filter order is a second set value, and a quantization bit number is a third set value, calculating an optimal solution of the transfer function of the noise; determining a value corresponding to the working variable parameters of the components according to the optimal solution; and selecting corresponding components for the circuit architecture based on the value, and generating a pulse density modulation interface of an analog-to-digital converter. The application can accurately design a PDM interface, and improves the signal processing performance of the interface.

The application belongs to the technical field of radar signal processing and imaging, and provides a differential coupling single-frequency time-varying low-quantization imaging method, which comprises receiving a SAR echo signal, constructing a single-frequency time-varying threshold signal, single-bit quantization processing of a reference channel, single-bit quantization processing of a phase shift channel, double-channel data fusion and harmonic cancellation, and imaging processing based on a synthesized signal. The application effectively reduces the quantization noise base and improves the detection capability of a SAR system on weak targets under the premise of not significantly increasing the hardware cost.

The present invention discloses a DOA estimation method for a single-bit sensor array, which relates to the field of signal processing technology, including obtaining an original signal, performing single-bit quantization on the original signal to obtain a quantization result, and calculating the covariance matrix of the quantization result; vectorizing the covariance matrix to generate a column vector; performing DFT operation on the column vector to obtain spectrum information; obtaining the peak value and the peak position of the spectrum information through a findpeaks function, processing to obtain the arrival angle position, and finding the left and right boundary positions of the arrival angle position to determine the arrival angle interval; using DTFT to perform frequency refinement on the arrival angle interval, and performing spectrum peak search on the refined arrival angle interval, calculating and obtaining the final angle estimation. The present invention has low computational complexity, and the estimated angle is more accurate than the same type of algorithms, and can be extended to other nonlinear array structures.

The embodiment of the present application discloses a method, device, equipment and medium for determining the spatial power characteristics of a signal, which relates to the field of signal processing technology, including: determining an array response vector based on the geometric structure of an antenna array; calculating a covariance matrix based on array received data obtained after sampling and quantization, and performing eigenvalue decomposition on the covariance matrix to obtain a noise subspace; establishing a first spatial spectrum function based on the noise subspace, and establishing a second spatial spectrum function based on the covariance matrix and the array response vector; performing spectrum peak search on the first spatial spectrum function and the second spatial spectrum function based on a preset angle interval to obtain a first azimuth, a second azimuth and a power of each interference signal; performing matching processing on the first azimuth and the second azimuth to obtain a matching result, and determining the spatial power characteristic information of each interference signal according to the matching result and the power of each interference signal, thereby improving the accuracy of determining the direction of arrival and the power of the signal.

The present invention belongs to the field of radio signal processing technology, and specifically relates to an infinite amplitude line spectrum estimation method based on robust compressed sensing and analog sampling ADC analog sampling when the amplitude of the signal arriving at the front end of the receiver exceeds the dynamic range of the ordinary ADC input. The analog sampling ADC analog sampling mode of the present invention and the proposed line spectrum estimation method utilize the robust compressed sensing theory to realize infinite line spectrum estimation when the amplitude of the signal arriving at the front end of the receiver is higher than the dynamic range of the ordinary ADC input, accurately estimate the amplitude of the signal and its carrier frequency information, and can overcome the problem of weak signal loss caused by direct sampling and quantization using an ordinary ADC. Experiments show that the present invention successfully estimates the amplitude and carrier frequency of multiple source signals when the input signal amplitude is higher than the ADC input dynamic range, thereby achieving reliable weak signal estimation performance at a low quantization level.

The application discloses a jitter auxiliary measurement precision enhancement method for space-based calibration, belongs to the technical field of electronic communication and signal processing, and comprises the following steps: receiving a continuous-time analog input signal from a space-based calibration system; injecting an artificial jitter signal with zero mean and subject to uniform distribution into the continuous-time analog input signal to generate a composite analog signal; performing multiple independent quantization observations on the composite analog signal by using a low-bit analog-to-digital converter, obtaining multiple quantization output values corresponding to discrete levels, calculating the arithmetic mean of the multiple quantization output values, and taking the arithmetic mean as a high-precision estimation value of the amplitude of the continuous-time analog input signal; and the jitter auxiliary measurement precision enhancement method for space-based calibration can greatly improve the measurement precision of the space-based calibration system without increasing the hardware power consumption and volume.

The invention discloses a step error-based full-electronic interlocking system turnout slab sampling method and device, and relates to the technical field of rail transit, and the method comprises the steps: carrying out the sampling of a turnout slab circuit AC signal based on a sampling frequency, and obtaining a discrete signal sequence; fitting sampling points of the discrete signal sequence according to discrete sine waves, and determining proportionality coefficients of the sampling points; calculating a weighted signal error of the sampling point according to the proportionality coefficient; comparing the weighted signal error with an error threshold value, and judging whether the sampled representation signal accords with a turnout in-place representation waveform or not according to a comparison result; the probability of misjudgment caused by rough signal quantization and misjudgment caused by interference points are reduced, the signal quality and the signal processing efficiency of sampling points are remarkably improved, the accuracy and the reliability of sampling representation are further remarkably improved, and the defects that due to inconsistent signal conversion and unstable signal quality, the signal quality is not stable, and the signal quality is not stable are overcome. Therefore, the problem of inaccurate and unreliable turnout state judgment is solved.

The invention discloses an asynchronous chaotic secret communication method and a system thereof. The transmitter obtains an estimated power threshold value according to system response of the receiver and preset calibration signal processing, and then two mode mapping sets are established; the transmitter performs self-synchronization coding on the original binary communication data to obtain coded binary communication data, and then performs encryption mapping on the coded binary communication data according to the two mode mapping sets to obtain a frequency spectrum mode sequence; the transmitter transmits a calibration signal, and the receiver sets a power threshold value according to the calibration signal; the transmitter transmits multi-mode radio frequency chaotic communication signals in different modes according to a frequency spectrum mode sequence, the receiver receives and carries out conditioning quantization to obtain recovered coded binary communication data, and then the recovered coded binary communication data are decoded to obtain recovered original binary communication data. The method is high in safety, strong in anti-interference capability, simple in system structure, low in cost, free of chaos synchronization, good in engineering realizability and capable of being used for secret communication in a complex electromagnetic environment.

The present invention relates to the field of data processing technology, and in particular to an audio signal processing method and system. The method comprises the following steps: acoustically simulating site plan data to generate acoustic topology data; collecting sound signals according to the acoustic topology data to generate multi-directional audio data; quantizing the multi-directional audio data to generate spectrum gradient matrix data; selecting high entropy data from the spectrum gradient matrix data to generate key feature data; optimizing the model based on the key feature data to generate a spatial audio reconstruction model; iteratively optimizing the spatial audio reconstruction model according to the site dynamic change parameters to generate a dynamic audio reconstruction model; collecting feedback data using sensor equipment to generate real-time feedback data; performing feedback control according to the real-time feedback data to generate sound propagation optimization data. The present invention improves the real-time transmission quality of audio signals by analyzing the sound field characteristics and environmental changes.

This invention discloses a method for covert processing of spaceborne signals based on Tent mapping, relating to the field of signal processing. In this invention, the transmitter generates a chaotic Tent sequence, which is then transformed by Box-Muller to obtain a Gaussian-distributed first interference signal. The chaotic sequence and the first interference signal are equally divided, and an identification signal containing a synchronization flag and a checksum is constructed based on the quantization value of the first data point of each chaotic sequence segment. An amplitude adjustment coefficient is determined for each segment of the target signal, and the adjusted first interference signal of each segment is superimposed with the corresponding segment of the target signal to obtain a mixed subsequence. An identification signal is added before each mixed subsequence to form a relay signal. At the receiver, a sliding window is used to match the synchronization flag in the received signal to locate the identification signal, the quantization value at the beginning of each segment is parsed, and a second interference subsequence is independently reconstructed. A coarse estimate is obtained by performing a two-dimensional correlation between time delay and frequency offset, and the received target signal is obtained through interference subspace projection cancellation. The method proposed in this invention has low hardware overhead, strong anti-interception capability, and high cancellation accuracy.

The invention discloses a lightweight lossless audio encoding and decoding method and system, and belongs to the technical field of audio signal processing. The method comprises the following steps: converting an original audio waveform into a compact coding expression through an encoder formed by connecting a TConv unit, a convolution unit, a down-sampling unit and a Local-Transform unit in series; a single finite scalar quantizer is adopted to discretize the coded representation, and a mixed quantization strategy is introduced to generate a single and consistent discrete token stream; and finally, a high-fidelity audio waveform is reconstructed through a decoder formed by connecting a Local-Transform unit, a TEConv unit, an up-sampling unit and an audio recovery unit in series. Through unit structure optimization and joint loss function training, indexes such as SDR, PESQ, STOI, WER and the like under multiple different code rates are all superior to those of a baseline model, model parameters are only 10.31-11.29 M, collaboration of lightweight deployment and high-fidelity reconstruction is achieved, resource-constrained equipment is adapted to downstream scenes such as voice communication, ASR and the like, and good engineering practicability is achieved.

The invention relates to the field of MIMO signal processing, and discloses an MIMO-oriented bit width quantization method, device, equipment and medium, and the method comprises the steps: obtaining a plurality of variables to be quantized, carrying out the quantization of the integer bit width and sign bit width of the variables, obtaining the initial population of the variables, and carrying out the iterative optimization of the initial population, in any iteration process, carrying out the quantization of the integer bit width and sign bit width of the variables, and carrying out the quantization of the integer bit width and sign bit width of the variables. The method comprises the steps of determining a reference individual in a current population, performing decimal bit width updating on other individuals based on the reference individual, performing decimal bit width updating on any individual after the decimal bit width updating based on the target function value difference between the individual and the other individual, and determining a final population after iterative optimization is completed. And determining a target individual in the final population, and quantizing the decimal bit widths of the variables based on the decimal bit width combination of the target individual. According to the technical scheme provided by the invention, the automatic and efficient distribution of the decimal bit width in the MIMO detector is realized, and the calculation complexity and the hardware overhead are reduced.

The invention provides a single-bit radar quantization method and system based on a frequency modulation threshold, and belongs to the technical field of signal processing. Comprising the steps that a master control MCU controls a DAC to generate a sawtooth signal and further controls a voltage-controlled oscillator to generate a frequency-modulated continuous wave, and a linear frequency-modulated signal is transmitted by a transmitting antenna after the frequency-modulated continuous wave is amplified by a power amplifier. The receiving antenna obtains an echo signal reflected by a target, the echo signal is amplified by the low-noise amplifier, the echo signal and the linear frequency modulation signal are mixed in the mixer to extract an intermediate frequency signal, and a single-frequency intermediate frequency signal is obtained through AGC automatic gain amplification and amplitude limiting. The main control MCU generates a frequency modulation threshold signal through a DAC or an oscillator, the frequency modulation threshold signal and the frequency modulation threshold signal are input into a voltage comparator for amplitude comparison to generate a binary signal, single-bit quantized data are obtained through high-speed sampling of a digital IO port, and finally quantization is completed through frequency domain transformation and rectangular window function filtering. According to the method, the problems of harmonic interference, false target influence and insufficient reservation of signal amplitude and phase information in the prior art are solved.

The application discloses an ADC based on a neural network enhanced heterogeneous channel architecture and relates to the technical field of ADC design.The ADC comprises two-way heterogeneous ADC channels, a delay module, a neural network circuit and a streaming training controller.The two-way heterogeneous ADC channels are a high-speed low-precision channel and a low-speed high-precision channel, and the two channels are connected to the same signal source.The delay module performs delay and alignment processing on the output data of the two-way channels, the multi-time dimension data formed after the high-speed channel is delayed is aligned with the single time dimension data of the low-speed channel, the high-speed channel data is taken as training data, the low-speed channel data is taken as label data, the neural network circuit is used for training to generate high-speed high-precision quantization output, and the streaming training controller globally manages data acquisition and batch processing operations and generates training data through downsampling, time offset and batch processing.The application can realize the combination of high speed and high precision and meet the complex signal processing requirement.

The application discloses a controllable sparse transfer attack method based on a generator, and belongs to the cross field of communication signal processing and artificial intelligence. The method aims to solve the problems of existing adversarial attacks, such as too concentrated disturbance in the electromagnetic signal field, insufficient concealment and difficulty in controlling the disturbance density. The technical scheme comprises the following steps: a deep generation network is used to simultaneously output a disturbance amplitude and a disturbance probability; a group mask matrix is generated by locating a region critical to classification through calculation of group spectral saliency; a binary position matrix is generated through a mask quantization network in combination with the group mask and the disturbance probability; a total loss function containing an adversarial loss and a smoothing loss is used to train the generation network; and finally, a sparse controllable adversarial sample is generated through the trained network. The application can generate an adversarial sample with strong concealment, high transferability and controllable disturbance proportion, and provides an effective means for security evaluation of a deep neural network.

The invention relates to the technical field of signal processing, and discloses a small signal amplification and measurement device based on a band-pass filter. The multi-stage band-pass filter comprises a multi-stage band-pass filter module, a pre-amplification module, an analog-digital conversion module and the like. The center frequency of the multi-stage band-pass filter is distributed according to logarithm intervals, the bandwidth is dynamically adjusted, the pre-amplification module realizes low-noise amplification, and the analog-to-digital conversion module adopts an oversampling technology and a Sigma-Delta modulator to improve the quantization resolution. The dynamic gain control module adjusts gain in real time, and the noise suppression module eliminates periodic interference. In addition, the device also has the functions of multi-sensor fusion, dynamic range expansion, signal integrity detection, automatic calibration, temperature compensation, parallel processing and the like. The device can accurately amplify and measure small signals and effectively suppress noise, is suitable for the fields of biomedicine, communication, industrial monitoring and the like, and overcomes the defects of a traditional device in the aspect of small signal processing.