Single-Channel Multiple-Receiver Sound Source Localization System with Homomorphic Deconvolution and Linear Regression

Abstract

The conventional sound source localization systems require the significant complexity because of multiple synchronized analog-to-digital conversion channels as well as the scalable algorithms. This paper proposes a single-channel sound localization system for transport with multiple receivers. The individual receivers are connected by the single analog microphone network which provides the superimposed signal over simple connectivity based on asynchronized analog circuit. The proposed system consists of two computational stages as homomorphic deconvolution and machine learning stage. A previous study has verified the performance of time-of-flight estimation by utilizing the non-parametric and parametric homomorphic deconvolution algorithms. This paper employs the linear regression with supervised learning for angle-of-arrival prediction. Among the circular configurations of receiver positions, the optimal location is selected for three-receiver structure based on the extensive simulations. The non-parametric method presents the consistent performance and Yule–Walker parametric algorithm indicates the least accuracy. The Steiglitz–McBride parametric algorithm delivers the best predictions with reduced model order as well as other parameter values. The experiments in the anechoic chamber demonstrate the accurate predictions in proper ensemble length and model order.

Gaussian Process Regression for Single-Channel Sound Source Localization System Based on Homomorphic Deconvolution

“…This section is presented by summarizing the previous papers [ 30 , 42 ] to provide access to the reader while maintaining concept integrity. For further information, the readers are suggested to follow the article for finding detailed explanation about the nonparametric and parameter HD.…”

“…The FFT (stage ①) and inverse FFT (stage ④) pair with the absolute logarithm (stage ②), which corresponds to the real cepstrum of the forward conversion of the homomorphic system. The previous paper [ 42 ] showed that the ensemble average (stage ③) after the logarithm (stage ②) denotes the inverse proportionality in noise variance. Therefore, the longer ensemble average length induces higher SNR due to the lower noise power.…”

“…This process significantly reduces the computation of evaluating the receiver configuration. The previous paper [ 42 ] performed the 680-combination evaluation, the prediction error analysis, on 14 computers (intel i7-7700/32GB memory) with 25-h execution time. This paper assesses the 1,254,890 combinations of the single computer (inter i7-10700k/32GB memory) for less than half hour.…”

“…Additionally, the computation and communication complexity should be maintained at a low level for system feasibility. The single-channel multiple-receiver SSL was suggested in the previous papers [ 30 , 42 ] to challenge the binaural/monaural method without the external structure and to include the beamforming performance with reduced computation/communication burden. The paper established the promising performance with a simple machine learning algorithm, linear regression.…”

Sound Source Localization Using a Convolutional Neural Network and Regression Model

“…For example, the frequency of sound emitted from different locations can be distinguished efficiently, and multiple sound sources can be located at the same time. Therefore, sound source localization (SSL) has attracted much attention in recent years [ 16 , 17 , 18 ].…”