Lu, Chris Xiaoxuan, Wen, Hongkai, Wang, Sen, Markham, Andrew and Trigoni, Niki (2017) SCAN : learning speaker identity from noisy sensor data. In: 16th International Conference on Information Processing in Sensor Networks, Pittsburgh, Pennsylvania, USA, 18-21 Apr 2017 . Published in: 16th ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN), 2017 pp. 67-78. ISBN 9781450348904 . doi:10.1145/3055031.3055073

Preview

PDF WRAP-SCAN-learning- speaker- identity- noisy-sensor- data-Wen-2017.pdf - Accepted Version - Requires a PDF viewer. Download (1503Kb) | Preview

Request Changes to record.

Abstract

Sensor data acquired from multiple sensors simultaneously is featuring increasingly in our evermore pervasive world. Buildings can be made smarter and more efficient, spaces more responsive to users. A fundamental building block towards smart spaces is the ability to understand who is present in a certain area. A ubiquitous way of detecting this is to exploit the unique vocal features as people interact with one another. As an example, consider audio features sampled during a meeting, yielding a noisy set of possible voiceprints. With a number of meetings and knowledge of participation (e.g. through a calendar or MAC address), can we learn to associate a specific identity with a particular voiceprint? Obviously enrolling users into a biometric database is time-consuming and not robust to vocal deviations over time. To address this problem, the standard approach is to perform a clustering step (e.g. of audio data) followed by a data association step, when identity-rich sensor data is available. In this paper we show that this approach is not robust to noise in either type of sensor stream; to tackle this issue we propose a novel algorithm that jointly optimises the clustering and association process yielding up to three times higher identification precision than approaches that execute these steps sequentially. We demonstrate the performance benefits of our approach in two case studies, one with acoustic and MAC datasets that we collected from meetings in a non-residential building, and another from an online dataset from recorded radio interviews.

Item Type:	Conference Item (Paper)
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering
Divisions:	Faculty of Science > Computer Science
Library of Congress Subject Headings (LCSH):	Sensor networks, Voiceprints -- Industrial applications
Journal or Publication Title:	16th ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN), 2017
Publisher:	ACM/IEEE
ISBN:	9781450348904
Official Date:	12 June 2017
Dates:	Date Event 12 June 2017 Available 18 January 2017 Accepted
Date of first compliant deposit:	23 March 2017
Page Range:	pp. 67-78
DOI:	10.1145/3055031.3055073
Status:	Peer Reviewed
Publication Status:	Published
Funder:	Engineering and Physical Sciences Research Council (EPSRC), University of Oxford. Department of Engineering Science, Google (Firm). Deep Mind
Grant number:	EP/J012017/1, EP/M017583/1, EP/M019918/1 (EPSRC)
RIOXX Funder/Project Grant:	Project/Grant ID RIOXX Funder Name Funder ID EP/J012017/1 [EPSRC] Engineering and Physical Sciences Research Council http://dx.doi.org/10.13039/501100000266 EP/M017583/1 [EPSRC] Engineering and Physical Sciences Research Council http://dx.doi.org/10.13039/501100000266 EP/M019918/1 [EPSRC] Engineering and Physical Sciences Research Council http://dx.doi.org/10.13039/501100000266 UNSPECIFIED University of Oxford. Department of Engineering Science UNSPECIFIED UNSPECIFIED Google http://dx.doi.org/10.13039/100006785
Conference Paper Type:	Paper
Title of Event:	16th International Conference on Information Processing in Sensor Networks
Type of Event:	Conference
Location of Event:	Pittsburgh, Pennsylvania, USA
Date(s) of Event:	18-21 Apr 2017
Related URLs:	Organisation Publisher

Request changes or add full text files to a record

Repository staff actions (login required)

View Item

Downloads