Binaural source localization using deep learning and head rotation information

Guillermo Garcia Barrios, Daniel Aleksander Krause, Archontis Politis, Annamaria Mesaros, Juana M. Gutierrez-Arriola, Ruben Fraile

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

3 Downloads (Pure)

Abstract

This work studies learning-based binaural sound source localization, under the influence of head rotation in reverberant conditions. Emphasis is on whether knowledge of head rotation can improve localization performance over the non-rotating case for the same acoustic scene. Simulations of binaural head signals of a static and rotating head were conducted, for 5 different rotation speeds and a wide range of reverberant conditions. Several convolutional recurrent neural network models were evaluated including a static head scenario, a model without rotation information, and distinct models differentiated on the way of manipulating the quaternions. The results were analyzed based on the direction-of-arrival error, and they show the importance of using quaternions as additional features, with the best localization accuracy obtained when using an additional convolutional branch that merges the features through addition or concatenation. Nevertheless, raw quaternion features presented lower performance than the static baseline model. Additionally, the study shows the importance of the analysis time window length when using information about head rotation.
Original languageEnglish
Title of host publication2022 30th European Signal Processing Conference (EUSIPCO)
Pages36-40
Number of pages5
ISBN (Electronic)978-90-827970-9-1
DOIs
Publication statusPublished - 18 Oct 2022
Publication typeA4 Article in conference proceedings
Event European Signal Processing Conference - Belgrade, Serbia
Duration: 29 Aug 20222 Sept 2022

Publication series

NameEuropean Signal Processing Conference
ISSN (Electronic)2076-1465

Conference

Conference European Signal Processing Conference
Country/TerritorySerbia
CityBelgrade
Period29/08/222/09/22

Publication forum classification

  • Publication forum level 1

Fingerprint

Dive into the research topics of 'Binaural source localization using deep learning and head rotation information'. Together they form a unique fingerprint.

Cite this