scholarly journals Machine listening in spatial acoustic scenes with deep networks in different microphone geometries

2020 ◽  
Vol 1 ◽  
pp. 6
Author(s):  
Jörn Anemüller
Keyword(s):  
Source Localization ◽  
Sound Field ◽  
Support Vector ◽  
Specific Information ◽  
Acoustic Source ◽  
Linear Classifier ◽  
Linear Approach ◽  
Machine Listening ◽  
Acoustic Source Localization ◽  
Multi Speech

Multi-channel acoustic source localization evaluates direction-dependentinter-microphone differences in order to estimate the position of an acousticsource embedded in an interfering sound field. We here investigate a deep neuralnetwork (DNN) approach to source localization that improves on previous workwith learned, linear support-vector-machine localizers. DNNs with depthsbetween 4 and 15 layers were trained to predict azimuth direction of targetspeech in 72 directional bins of width 5 degree, embedded in an isotropic,multi-speech-source noise field. Several system parameters were varied, inparticular number of microphones in the bilateral hearing aid scenario wasset to 2, 4, and 6, respectively. Results show that DNNs provide a clear improvement inlocalization performance over a linear classifier reference system.Increasing the number of microphones from 2 to 4 results in a larger increase ofperformance for the DNNs than for the linear system. However, 6 microphonesprovide only a small additional gain. The DNN architectures perform betterwith 4 microphones than the linear approach does with 6 microphones, thusindicating that location-specific information in source-interference scenariosis encoded non-linearly in the sound field.

Novel Acoustic Source Localization Method in WSN Based on LSSVR Regression Modeling

2012 ◽  
Vol 468-471 ◽  
pp. 2296-2303
Author(s):  
Xiao Ping Zhang ◽  
Yang Wang
Keyword(s):  
Source Localization ◽  
Feature Vector ◽  
Signal To Noise Ratio ◽  
Parameters Optimization ◽  
Support Vector ◽  
Acoustic Source ◽  
Signal To Noise ◽  
Localization Method ◽  
Acoustic Source Localization ◽  
Noise Ratio

To solve the problem of acoustic source localization in wireless sensor networks (WSN) under interference of environmental noise, a novel acoustic source localization method in WSN based on Least Square Support Vector Regression (LSSVR) modeling (ASL-LRM) was proposed. The ideal measured values of acoustic sensors were used to compose feature vector at first. Then LSSVR models were built by LSSVR modeling on the mapping relation between feature vector and acoustic source coordinate. The acoustic source was then located by inputting feature vector composed of real measured values of the sensors into LSSVR models. The modeling parameters optimization method based on localization effect in sample locations was also discussed. Experiments were performed in 100 test locations. RMSE values by ASL-LRM method in 72-76 test locations were less than MLE method and reduced by 60%-74% at most. In lower signal-to-noise ratio case, there were 87 test locations where RMSE values by ASL-LRM method were less than 2 meters, while there were only 12 test locations by MLE method. It shows ASL-LRM method achieves better localization effects in a large part of the region surrounded by sensor nodes. It especially has advantage on the occasions like lower signal-to-noise ratio or high precision localization.

Utilizing laser distance sensors for audio recovery and acoustic source localization

2021 ◽  
Vol 149 (4) ◽  
pp. A43-A44
Author(s):  
Kanad Sarkar ◽  
Ryan M. Corey ◽  
Sri Vuppala ◽  
Andrew C. Singer
Keyword(s):  
Source Localization ◽  
Acoustic Source ◽  
Acoustic Source Localization

A fast approach for acoustic source localization on a thin spherical shell

2021 ◽  
pp. 147592172110419
Author(s):  
Zixian Zhou ◽  
Zhiwen Cui ◽  
Tribikram Kundu
Keyword(s):  
Velocity Profile ◽  
Spherical Shell ◽  
Source Localization ◽  
Pressure Vessels ◽  
Solution Technique ◽  
Acoustic Source ◽  
Spherical Shells ◽  
Fast Approach ◽  
Acoustic Source Localization ◽  
Thin Spherical Shell

Thin spherical shell structures are wildly used as pressure vessels in the industry because of their property of having equal in-plane normal stresses in all directions. Since very large pressure difference between the inside and outside of the wall exists, any formation of defects in the pressure vessel wall has a huge safety risk. Therefore, it is necessary to quickly locate the area where the defect maybe located in the early stage of defect formation and make repair on time. The conventional acoustic source localization techniques for spherical shells require either direction-dependent velocity profile knowledge or a large number of sensors to form an array. In this study, we propose a fast approach for acoustic source localization on thin isotropic and anisotropic spherical shells. A solution technique based on the time difference of arrival on a thin spherical shell without the prior knowledge of direction-dependent velocity profile is provided. With the help of “L”-shaped sensor clusters, only 6 sensors are required to quickly predict the acoustic source location for anisotropic spherical shells. For isotropic spherical shells, only 4 sensors are required. Simulation and experimental results show that this technique works well for both isotropic and anisotropic spherical shells.

Sign in / Sign up

Export Citation Format

Share Document