Learning from data in the quaternion domain enables us to exploit internal dependencies of 4D signals and treating them as a single entity. One of the models that perfectly suits with quaternion-valued data processing is represented by 3D acoustic signals in their spherical harmonics decomposition. In this paper, we address the problem of localizing and detecting sound events in the spatial sound field by using quaternion-valued data processing. In particular, we consider the spherical harmonic components of the signals captured by a first-order ambisonic microphone and process them by using a quaternion convolutional neural network. Experimental results show that the proposed approach exploits the correlated nature of the ambisonic signals, thus improving accuracy results in 3D sound event detection and localization.

Quaternion convolutional neural networks for detection and localization of 3D sound events / Comminiello, D; Lella, M; Scardapane, S; Uncini, A. - 2019:(2019), pp. 8533-8537. (Intervento presentato al convegno IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) tenutosi a Brighton; United Kingdom) [10.1109/ICASSP.2019.8682711].

Quaternion convolutional neural networks for detection and localization of 3D sound events

Comminiello, D;Scardapane, S;Uncini, A
2019

Abstract

Learning from data in the quaternion domain enables us to exploit internal dependencies of 4D signals and treating them as a single entity. One of the models that perfectly suits with quaternion-valued data processing is represented by 3D acoustic signals in their spherical harmonics decomposition. In this paper, we address the problem of localizing and detecting sound events in the spatial sound field by using quaternion-valued data processing. In particular, we consider the spherical harmonic components of the signals captured by a first-order ambisonic microphone and process them by using a quaternion convolutional neural network. Experimental results show that the proposed approach exploits the correlated nature of the ambisonic signals, thus improving accuracy results in 3D sound event detection and localization.
2019
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
quaternion neural networks; hypercomplex machine learning; 3D audio; ambisonics
04 Pubblicazione in atti di convegno::04b Atto di convegno in volume
Quaternion convolutional neural networks for detection and localization of 3D sound events / Comminiello, D; Lella, M; Scardapane, S; Uncini, A. - 2019:(2019), pp. 8533-8537. (Intervento presentato al convegno IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) tenutosi a Brighton; United Kingdom) [10.1109/ICASSP.2019.8682711].
File allegati a questo prodotto
File Dimensione Formato  
Comminiello_ Quaternion-convolutional_2019.pdf

solo gestori archivio

Tipologia: Versione editoriale (versione pubblicata con il layout dell'editore)
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 195.71 kB
Formato Adobe PDF
195.71 kB Adobe PDF   Contatta l'autore

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11573/1335709
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 37
  • ???jsp.display-item.citation.isi??? 31
social impact