Localization, Detection and Tracking of Multiple Moving Sound Sources with a Convolutional Recurrent Neural Network

Sharath Adavanne; Archontis Politis; Tuomas Virtanen

Localization, Detection and Tracking of Multiple Moving Sound Sources with a Convolutional Recurrent Neural Network

Sharath Adavanne
, Archontis Politis
, Tuomas Virtanen

Computing Sciences

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

Abstract

This paper investigates the joint localization, detection, and tracking of sound events using a convolutional recurrent neural network (CRNN). We use a CRNN previously proposed for the localization and detection of stationary sources, and show that the recurrent layers enable the spatial tracking of moving sources when trained with dynamic scenes. The tracking performance of the CRNN is compared with a stand-alone tracking method that combines a multi-source (DOA) estimator and a particle filter. Their respective performance is evaluated in various acoustic conditions such as anechoic and reverberant scenarios, stationary and moving sources at several angular velocities, and with a varying number of overlapping sources. The results show that the CRNN manages to track multiple sources more consistently than the parametric method across acoustic scenarios, but at the cost of higher localization error.

Original language	English
Title of host publication	Proceedings of the Detection and Classification of Acoustic Scenes and Events 2019 Workshop (DCASE2019)
Pages	20-24
ISBN (Electronic)	978-0-578-59596-2
Publication status	Published - Oct 2019
Publication type	A4 Article in conference proceedings
Event	Workshop on Detection and Classification of Acoustic Scenes and Events - New York, United States Duration: 25 Oct 2019 → 26 Oct 2019

Workshop

Workshop	Workshop on Detection and Classification of Acoustic Scenes and Events
Abbreviated title	DCASE
Country/Territory	United States
City	New York
Period	25/10/19 → 26/10/19

Publication forum classification

No publication forum level

Access to Document

http://dcase.community/documents/workshop2019/proceedings/DCASE2019Workshop_Adavanne_46.pdf

TAU-NIGENS Spatial Sound Events 2020
Politis, A. (Creator), Adavanne, S. (Creator) & Virtanen, T. (Creator), Zenodo, 31 May 2020
DOI: 10.5281/zenodo.4064792
Dataset

Cite this

Adavanne, S., Politis, A., & Virtanen, T. (2019). Localization, Detection and Tracking of Multiple Moving Sound Sources with a Convolutional Recurrent Neural Network. In Proceedings of the Detection and Classification of Acoustic Scenes and Events 2019 Workshop (DCASE2019) (pp. 20-24) http://dcase.community/documents/workshop2019/proceedings/DCASE2019Workshop_Adavanne_46.pdf

@inproceedings{559ea0de20eb4365887e321ce70a3b9a,

title = "Localization, Detection and Tracking of Multiple Moving Sound Sources with a Convolutional Recurrent Neural Network",

abstract = "This paper investigates the joint localization, detection, and tracking of sound events using a convolutional recurrent neural network (CRNN). We use a CRNN previously proposed for the localization and detection of stationary sources, and show that the recurrent layers enable the spatial tracking of moving sources when trained with dynamic scenes. The tracking performance of the CRNN is compared with a stand-alone tracking method that combines a multi-source (DOA) estimator and a particle filter. Their respective performance is evaluated in various acoustic conditions such as anechoic and reverberant scenarios, stationary and moving sources at several angular velocities, and with a varying number of overlapping sources. The results show that the CRNN manages to track multiple sources more consistently than the parametric method across acoustic scenarios, but at the cost of higher localization error.",

author = "Sharath Adavanne and Archontis Politis and Tuomas Virtanen",

year = "2019",

month = oct,

language = "English",

pages = "20--24",

booktitle = "Proceedings of the Detection and Classification of Acoustic Scenes and Events 2019 Workshop (DCASE2019)",

note = "Workshop on Detection and Classification of Acoustic Scenes and Events, DCASE ; Conference date: 25-10-2019 Through 26-10-2019",

}

Adavanne, S, Politis, A & Virtanen, T 2019, Localization, Detection and Tracking of Multiple Moving Sound Sources with a Convolutional Recurrent Neural Network. in Proceedings of the Detection and Classification of Acoustic Scenes and Events 2019 Workshop (DCASE2019). pp. 20-24, Workshop on Detection and Classification of Acoustic Scenes and Events, New York, United States, 25/10/19. <http://dcase.community/documents/workshop2019/proceedings/DCASE2019Workshop_Adavanne_46.pdf>

Localization, Detection and Tracking of Multiple Moving Sound Sources with a Convolutional Recurrent Neural Network. / Adavanne, Sharath; Politis, Archontis ; Virtanen, Tuomas.
Proceedings of the Detection and Classification of Acoustic Scenes and Events 2019 Workshop (DCASE2019). 2019. p. 20-24.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Localization, Detection and Tracking of Multiple Moving Sound Sources with a Convolutional Recurrent Neural Network

AU - Adavanne, Sharath

AU - Politis, Archontis

AU - Virtanen, Tuomas

PY - 2019/10

Y1 - 2019/10

N2 - This paper investigates the joint localization, detection, and tracking of sound events using a convolutional recurrent neural network (CRNN). We use a CRNN previously proposed for the localization and detection of stationary sources, and show that the recurrent layers enable the spatial tracking of moving sources when trained with dynamic scenes. The tracking performance of the CRNN is compared with a stand-alone tracking method that combines a multi-source (DOA) estimator and a particle filter. Their respective performance is evaluated in various acoustic conditions such as anechoic and reverberant scenarios, stationary and moving sources at several angular velocities, and with a varying number of overlapping sources. The results show that the CRNN manages to track multiple sources more consistently than the parametric method across acoustic scenarios, but at the cost of higher localization error.

AB - This paper investigates the joint localization, detection, and tracking of sound events using a convolutional recurrent neural network (CRNN). We use a CRNN previously proposed for the localization and detection of stationary sources, and show that the recurrent layers enable the spatial tracking of moving sources when trained with dynamic scenes. The tracking performance of the CRNN is compared with a stand-alone tracking method that combines a multi-source (DOA) estimator and a particle filter. Their respective performance is evaluated in various acoustic conditions such as anechoic and reverberant scenarios, stationary and moving sources at several angular velocities, and with a varying number of overlapping sources. The results show that the CRNN manages to track multiple sources more consistently than the parametric method across acoustic scenarios, but at the cost of higher localization error.

M3 - Conference contribution

SP - 20

EP - 24

BT - Proceedings of the Detection and Classification of Acoustic Scenes and Events 2019 Workshop (DCASE2019)

T2 - Workshop on Detection and Classification of Acoustic Scenes and Events

Y2 - 25 October 2019 through 26 October 2019

ER -

Localization, Detection and Tracking of Multiple Moving Sound Sources with a Convolutional Recurrent Neural Network

Abstract

Workshop

Publication forum classification

Access to Document

Fingerprint

Datasets

TAU-NIGENS Spatial Sound Events 2020

Cite this