k-SpecNET: Localization and classification of indoor superimposed sound for acoustic sensor networks

Wei  Wang; Fatjon  Seraj; Paul J.M. Havinga

doi:10.1109/PerComWorkshops48775.2020.9156174

k-SpecNET: Localization and classification of indoor superimposed sound for acoustic sensor networks

Wei Wang, Fatjon Seraj, Paul J.M. Havinga

Pervasive Systems

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

3 Citations (Scopus)

97 Downloads (Pure)

Abstract

Automatic localization and classification of environmental sound events can provide great aid to many human-centric applications. However as many papers have mentioned, environmental sound events in daily life are complicated and hard to classify especially when multiple sounds happen simultaneously. Being different from many other works, we use an acoustic-sensor-network to solve this problem and decompose overlapping sound events using a sound localization model. The core of our contribution is to first find and locate the keypoints from each microphone's spectrogram and then aggregate them. With these aggregated keypoints as input, we then use 2 different classification models to further classify the type of sound sources. Compared with other classification models that only use single microphone, our experiments show that our solution is both accurate and low-cost in terms of calculation effort.

Original language	English
Title of host publication	2020 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops)
Place of Publication	Piscataway, NJ
Publisher	IEEE
Number of pages	6
ISBN (Electronic)	978-1-7281-4716-1
ISBN (Print)	978-1-7281-4717-8
DOIs	https://doi.org/10.1109/PerComWorkshops48775.2020.9156174
Publication status	Published - 4 Aug 2020
Event	2020 IEEE International Conference on Pervasive Computing and Communications Workshops, PerCom 2020 - University of Austin, Austin, United States Duration: 23 Mar 2020 → 27 Mar 2020 Conference number: 18 http://percom.org/Previous/ST2020/

Conference

Conference	2020 IEEE International Conference on Pervasive Computing and Communications Workshops, PerCom 2020
Abbreviated title	PerCom
Country/Territory	United States
City	Austin
Period	23/03/20 → 27/03/20
Internet address	http://percom.org/Previous/ST2020/

Keywords

Convolutional Neural Networks
Deep Neural Network
Environmental sound localization
Keypoints localization
TDOA
acoustic sensors
superimposed sound
22/2 OA procedure

Access to Document

10.1109/PerComWorkshops48775.2020.9156174

Wang2020k-specnetFinal published version, 8.01 MBLicence: Taverne

Cite this

@inproceedings{6ef4715788b84af5a1bdbf61fe7000ce,

title = "k-SpecNET: Localization and classification of indoor superimposed sound for acoustic sensor networks",

abstract = "Automatic localization and classification of environmental sound events can provide great aid to many human-centric applications. However as many papers have mentioned, environmental sound events in daily life are complicated and hard to classify especially when multiple sounds happen simultaneously. Being different from many other works, we use an acoustic-sensor-network to solve this problem and decompose overlapping sound events using a sound localization model. The core of our contribution is to first find and locate the keypoints from each microphone's spectrogram and then aggregate them. With these aggregated keypoints as input, we then use 2 different classification models to further classify the type of sound sources. Compared with other classification models that only use single microphone, our experiments show that our solution is both accurate and low-cost in terms of calculation effort.",

keywords = "Convolutional Neural Networks, Deep Neural Network, Environmental sound localization, Keypoints localization, TDOA, acoustic sensors, superimposed sound, 22/2 OA procedure",

author = "Wei Wang and Fatjon Seraj and Havinga, {Paul J.M.}",

note = "Publisher Copyright: {\textcopyright} 2020 IEEE.; 2020 IEEE International Conference on Pervasive Computing and Communications Workshops, PerCom 2020, PerCom ; Conference date: 23-03-2020 Through 27-03-2020",

year = "2020",

month = aug,

day = "4",

doi = "10.1109/PerComWorkshops48775.2020.9156174",

language = "English",

isbn = "978-1-7281-4717-8",

booktitle = "2020 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops)",

publisher = "IEEE",

address = "United States",

url = "http://percom.org/Previous/ST2020/",

}

Wang, W, Seraj, F & Havinga, PJM 2020, k-SpecNET: Localization and classification of indoor superimposed sound for acoustic sensor networks. in 2020 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops)., 9156174, IEEE, Piscataway, NJ, 2020 IEEE International Conference on Pervasive Computing and Communications Workshops, PerCom 2020, Austin, Texas, United States, 23/03/20. https://doi.org/10.1109/PerComWorkshops48775.2020.9156174

k-SpecNET: Localization and classification of indoor superimposed sound for acoustic sensor networks. / Wang, Wei ; Seraj, Fatjon ; Havinga, Paul J.M.
2020 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops). Piscataway, NJ: IEEE, 2020. 9156174.

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

TY - GEN

T1 - k-SpecNET: Localization and classification of indoor superimposed sound for acoustic sensor networks

AU - Wang, Wei

AU - Seraj, Fatjon

AU - Havinga, Paul J.M.

N1 - Conference code: 18

PY - 2020/8/4

Y1 - 2020/8/4

N2 - Automatic localization and classification of environmental sound events can provide great aid to many human-centric applications. However as many papers have mentioned, environmental sound events in daily life are complicated and hard to classify especially when multiple sounds happen simultaneously. Being different from many other works, we use an acoustic-sensor-network to solve this problem and decompose overlapping sound events using a sound localization model. The core of our contribution is to first find and locate the keypoints from each microphone's spectrogram and then aggregate them. With these aggregated keypoints as input, we then use 2 different classification models to further classify the type of sound sources. Compared with other classification models that only use single microphone, our experiments show that our solution is both accurate and low-cost in terms of calculation effort.

AB - Automatic localization and classification of environmental sound events can provide great aid to many human-centric applications. However as many papers have mentioned, environmental sound events in daily life are complicated and hard to classify especially when multiple sounds happen simultaneously. Being different from many other works, we use an acoustic-sensor-network to solve this problem and decompose overlapping sound events using a sound localization model. The core of our contribution is to first find and locate the keypoints from each microphone's spectrogram and then aggregate them. With these aggregated keypoints as input, we then use 2 different classification models to further classify the type of sound sources. Compared with other classification models that only use single microphone, our experiments show that our solution is both accurate and low-cost in terms of calculation effort.

KW - Convolutional Neural Networks

KW - Deep Neural Network

KW - Environmental sound localization

KW - Keypoints localization

KW - TDOA

KW - acoustic sensors

KW - superimposed sound

KW - 22/2 OA procedure

UR - http://www.scopus.com/inward/record.url?scp=85092008806&partnerID=8YFLogxK

U2 - 10.1109/PerComWorkshops48775.2020.9156174

DO - 10.1109/PerComWorkshops48775.2020.9156174

M3 - Conference contribution

SN - 978-1-7281-4717-8

BT - 2020 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops)

PB - IEEE

CY - Piscataway, NJ

T2 - 2020 IEEE International Conference on Pervasive Computing and Communications Workshops, PerCom 2020

Y2 - 23 March 2020 through 27 March 2020

ER -

k-SpecNET: Localization and classification of indoor superimposed sound for acoustic sensor networks

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this