Signal-informed DNN-based DOA Estimation combining an External Microphone and GCC-PHAT Features

TL;DR

This paper introduces a signal-informed deep neural network approach for estimating the direction of arrival of a speaker in multi-talker environments, utilizing an external microphone and GCC-PHAT features to improve accuracy.

Contribution

It presents a novel method that applies a binary mask based on an external microphone's power distribution to enhance DOA estimation without prior knowledge of interfering speakers.

Findings

Improved localization accuracy in reverberant scenarios.

Effective use of external microphone data for DOA estimation.

No need for prior information about interfering speakers.

Abstract

Aiming at estimating the direction of arrival (DOA) of a desired speaker in a multi-talker environment using a microphone array, in this paper we propose a signal-informed method exploiting the availability of an external microphone attached to the desired speaker. The proposed method applies a binary mask to the GCC-PHAT input features of a convolutional neural network, where the binary mask is computed based on the power distribution of the external microphone signal. Experimental results for a reverberant scenario with up to four interfering speakers demonstrate that the signal-informed masking improves the localization accuracy, without requiring any knowledge about the interfering speakers.