Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A Method for Indoor Sound Source Area Localization Based on Convolutional Neural Network

A convolutional neural network and regional positioning technology, which is applied in the field of determining the location of signal sources by sound waves, can solve problems such as lack of adaptability and insufficient positioning accuracy

Active Publication Date: 2020-05-12
HEBEI UNIV OF TECH
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The technical problem to be solved by the present invention is to provide a method for locating indoor sound source areas based on convolutional neural networks. By converting sound source signals into spectrograms and inputting them into convolutional neural networks, the indoor single sound source can be realized. Regional positioning, overcome the lack of positioning accuracy and adaptability of the existing sound source localization technology in the unstructured indoor environment when the position of the sound source that people are interested in is only limited to some predefined areas lack of defects

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
  • A Method for Indoor Sound Source Area Localization Based on Convolutional Neural Network
  • A Method for Indoor Sound Source Area Localization Based on Convolutional Neural Network
  • A Method for Indoor Sound Source Area Localization Based on Convolutional Neural Network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0086] A method for locating an indoor sound source area based on a convolutional neural network in this embodiment, the specific steps are as follows:

[0087] The first step is to build a signal model:

[0088] The detailed process of establishing the signal model is that in an unstructured indoor environment, a single fixed sound source s(t) is set in a two-dimensional space. For an array composed of M=4 microphones, the i-th microphone receives The received sound signal is shown in the following formula (1):

[0089] x i (t) = α i s(t-τ i )+n i (t)i=1,2,...,M(1),

[0090] In formula (1), x i (t) represents the sound signal received by the i-th microphone, i represents the i-th microphone, α i and τ i respectively represent the amplitude attenuation factor and relative time delay of the sound signal received from the sound source, n i (t) is the sum of various noise signals. The sound signal and the noise signal received by each microphone are set to be independent...

Embodiment 2

[0131] This embodiment is to illustrate the feasibility and effectiveness of the designed convolutional neural network framework. The present invention uses the trained convolutional neural network model to predict the test samples through experimental simulation, and obtains the classification result, that is, the sound source belongs to. The location of the area, and visualize the final test results through the tensorboard tool, is to use the trained network model to predict 10% of the spectrogram test samples, and obtain the classification result, that is, the accuracy of the area where the sound source belongs. In order to illustrate the feasibility and effectiveness of the convolutional neural network framework designed, the present invention is tested by experimental simulation, and the signal-to-noise ratio is selected as 5db, 10db, and 15db, and the tests are performed respectively, and the training is performed five times, and the same parameters are used for training ...

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 provides a method for region positioning of an indoor sound source based on a convolutional neural network and relates to a technology for determination of a signal source by use of a sound wave. A sound source signal is converted into the form of sound spectrograph and is input into the convolutional neural network to implement region positioning of the indoor simple sound source. The method comprises the steps of establishing a signal model; selecting a data sample on the basis of the established signal model; performing time-frequency analysis on sound signals collected by microphones M0, M1, M2 and M3, and establishing a positioning database; and performing training of the convolutional neural network on the constructed positioning database and implementing the region positioning of the indoor simple sound source based on the convolutional neural network. With the method for region positioning of the indoor sound source based on the convolutional neural network, the defects that insufficient positioning precision and poor adaptability in an unstructured indoor environment emerge in an existing sound source positioning technology when a sound position interested bypeople is only limited in predefined regions are overcome.

Description

technical field [0001] The technical solution of the present invention relates to the technology of using sound waves to determine the location of signal sources, specifically a method for locating indoor sound source areas based on convolutional neural networks. Background technique [0002] The sound source localization technology based on microphone array is a research hotspot at home and abroad in recent years. The existing sound source localization methods based on microphone array can be roughly divided into three categories: controllable beamforming technology based on maximum output power, High-resolution spectral estimation technology and sound source localization technology based on time difference of arrival. Most of these existing methods are based on geometric models of sound propagation and energy attenuation. Due to the large influence of the environment and the high degree of model dependence, there are still certain limitations in applying them to unstructur...

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 Patents(China)
IPC IPC(8): G01S5/20
CPCG01S5/20
Inventor 孙昊张晓萌王硕朋徐静翟葆朔
Owner HEBEI UNIV OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com