Single-channel signal double-channel separation method and device, storage medium and processor

A separation method, single-channel technology, applied in neural learning methods, instruments, speech analysis, etc., can solve the unpredictability of noise randomness of single-channel mixed-signal data with noise, the relationship between target signal and data is not direct, and the difficulty of target signal data. Prediction range, etc.

Inactive Publication Date: 2019-10-11
SOUTH CHINA NORMAL UNIVERSITY
View PDF3 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0009] However, the single-channel noisy mixed-signal data has noise randomness and unpredictability, and the relationship with the target signal data is not direct and clear. The main disadvantages of this type of method are: ①The model estimation is difficult; ②The model training speed is slow; ③ Model generalization is poor
[0012] But in the real environment, the main disadvantages of this kind of method are: the range of the target signal data is difficult to predict, and there are often physical interference phenomena caused by the unequal phases of the target signal data and the noise signal data

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Single-channel signal double-channel separation method and device, storage medium and processor
  • Single-channel signal double-channel separation method and device, storage medium and processor
  • Single-channel signal double-channel separation method and device, storage medium and processor

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0041] For the data separation of single-channel signals, target mapping and time-frequency masking respectively used in the prior art have their own defects. Therefore, the present invention draws on the idea of ​​multi-channel neural networks to propose a single-channel signal that combines the above two methods Two-way separation method.

[0042]The main feature of the multi-channel neural network is that there are no multiple models, and no pre-training is required. Instead, multiple independent branches with different model structures and data processing logic are designed on the basis of a single model, and are often followed by each branch. Add a fully connected layer for confluence, and finally perform unified overall training of each branch of the entire model through the backpropagation BP algorithm. Its features are multi-branch, combined training, and branch merging. The main idea of ​​the multi-channel neural network is to multi-process single or multi-modal data...

Embodiment 2

[0118] This embodiment provides a dual-channel separation device for a single-channel signal, which corresponds to the separation method described in Embodiment 1. The device includes a multi-channel neural network learning model module, which includes a target mapping path, a time-frequency masking path and Fully connected layer, where:

[0119] A multi-channel neural network learning model module, which includes a target mapping channel, a time-frequency masking channel, and a fully connected layer, wherein:

[0120] The target mapping path is used to separate the single-channel signal data by using the target mapping method,

[0121] A time-frequency masking path is used to separate single-channel signal data by using a time-frequency masking method;

[0122] The fully connected layer merging module is used to combine the output data after the target mapping path and the time-frequency masking path are separated, and sort them into the specifications of the target data, an...

Embodiment 3

[0129] This embodiment provides a storage medium on which a computer program is stored. When the program is running, the time-series physiological data classification method described in Embodiment 1 can be executed.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a single-channel signal double-channel separation method and device, a storage medium and a processor. The method comprises the steps that a multi-path neural network learningmodel is established, the model comprises a target mapping path, a time-frequency masking path and a full connection layer, the target mapping path adopts a target mapping method to separate and concurrently, and the time-frequency masking path adopts a time-frequency masking method to separate single-channel signal data; and the data output after the target mapping path and the time-frequency masking path are separated are converged through a full connection layer and arranged into the specification of the target data, and then estimated target signal data characteristics are output. According to the method, the advantages of a time-frequency masking method and a target mapping method are compatible, the defects of the time-frequency masking method and the target mapping method are overcome to a certain extent, and the generalization performance of the model is good under the condition that the signal data phase is not considered.

Description

technical field [0001] The present invention belongs to the research field of blind source separation (Blind Source Separation, BSS), which is mainly used in the field of signal processing in the early stage, also known as blind signal separation, and particularly relates to a single-channel signal double-channel separation method, device, storage medium and processor . Background technique [0002] At present, most of them regard the separation process of signal data as a supervised learning problem, and then use the deep learning network model to realize it. The general framework of blind source separation based on deep learning is mainly divided into two stages: "deep learning model training" and "single-channel data separation": [0003] (1) Training stage: use the deep learning model to extract the characteristics of the training data, and learn the nonlinear relationship between the unseparated source signal data and the manually separated label signal data; [0004]...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/62G06N3/04G06N3/08G10L21/0264G10L21/0308G10L25/30
CPCG06N3/084G10L21/0264G10L21/0308G10L25/30G06N3/048G06N3/044G06N3/045G06F2218/08G06F2218/12G06F18/241
Inventor 聂瑞华高卓君梁志浩
Owner SOUTH CHINA NORMAL UNIVERSITY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap