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

Voice DOA estimation method based on ResNet

A DOA and voice technology, which is applied to the direction or offset system, direction finder using ultrasonic/sonic/infrasonic waves, etc., can solve the problem of inaccurate voice DOA estimation, and achieve the effect of reducing network complexity

Active Publication Date: 2019-03-19
NANJING UNIV OF INFORMATION SCI & TECH
View PDF6 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a ResNet-based voice DOA estimation method, which can effectively solve the problem of inaccurate voice DOA estimation under strong noise reverberation conditions, is a DOA estimation method suitable for arbitrary array structures

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
  • Voice DOA estimation method based on ResNet
  • Voice DOA estimation method based on ResNet
  • Voice DOA estimation method based on ResNet

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0056] Such as figure 1 As shown, the present invention provides a kind of speech DOA estimation method based on ResNet, extracts feature from generalized cross-correlation (GCC), utilizes ResNet to learn the nonlinear mapping relation between feature and DOA from a large amount of simulated microphone array signals, in On the basis of the rough estimation of the traditional wideband MUSIC method, multiple ResNets are used for accurate and robust DOA estimation.

[0057] The technical solutions of the present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0058] Broadband MUSIC positioning

[0059] The microphone array has M array elements with a spacing of d, and each array element is the same omnidirectional microphone, and the far-field signal is incident at an angle of θ. Assume that the noise is Gaussian white noise independent of the incident signal, with a mean of 0 and a variance of σ 2 , then th...

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 voice DOA estimation method based on ResNet, which comprises the following steps of: 1, simulating a training data set by using MATLAB, traversing a measurement range by using a plurality of voice signals in the data set, and storing corresponding angles and voice signals; 2, after each simulation signal is subjected to framing processing, calculating GCC and performing phase transformation; cutting according to the array model parameters, weighting and summing each voice frame; storing the weighted features and the corresponding incident angle as the data set; 3, initializing ResNet by using MATConvNet and training by using the data set; 4, carrying out coarse positioning on the signal to be measured by using broadband MUSIC to obtain a coarse positioning result,and selecting a group ResNet with a center point closest to the broadband MUSIC result to carry out subsequent accurate positioning according to the coarse positioning result to obtain a DOA estimation result. The method can effectively solve the problem of inaccurate voice DOA estimation under the condition of strong noise reverberation, and is a DOA estimation method suitable for any array structure.

Description

technical field [0001] The invention belongs to the technical field of microphone array DOA estimation, in particular to a ResNet-based speech DOA estimation method, which can realize precise positioning of speech under strong noise reverberation conditions. Background technique [0002] Direction of Arrival (Direction of Arrival) estimation is one of the important directions of array signal processing, and it is widely used in remote automatic speech recognition, teleconferencing and automatic camera steering. However, it is difficult to obtain an accurate DOA estimate when the signal is distorted by strong noise and room reverberation. Therefore, robust DOA estimation under indoor conditions is required. Traditional DOA estimation methods in noisy and reverberant environments can be mainly divided into: (1) subspace methods, such as multiple signal classification (MUSIC) and estimation of signal parameters with rotation invariant techniques (Esprit); (2) generalized mutua...

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
IPC IPC(8): G01S3/802
CPCG01S3/802
Inventor 郭业才张浩然顾弘毅
Owner NANJING UNIV OF INFORMATION SCI & 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