Non-linguistic Vocalisation Recognition Based on Hybrid GMM-SVM Approach

Artur Janicki

Abstract

This paper describes an algorithm for detection of onlinguistic vocalisations, such as laughter or fillers, based on acoustic features. The algorithm proposed combines the benefits of Gaussian mixture models (GMM) and the advantages of support vector machines (SVMs). Three GMMs were trained for garbage, laughter, and fillers, and then an SVM model was trained in the GMM score space. Various experiments were run to tune the parameters of the proposed algorithm, using the data sets originating from the SSPNet Vocalisation Corpus (SVC) provided for the Social Signals Sub-Challenge of the INTERSPEECH 2013 Computational Paralinguistics Challenge. The results showed a remarkable growth of the unweighted average of the area under the receiver operating curve (UAAUC) compared to the baseline results (from 87.6% to over 94% for the development set), which confirmed the efficiency of the proposed method.
Author Artur Janicki (FEIT / IT)
Artur Janicki,,
- The Institute of Telecommunications
Pages153-157
Publication size in sheets0.5
Book Bimbot Frédéric, Fougeron Cécile, Pellegrino François, Cerisara Christophe (eds.): Proceedings of 14th Annual Conference of the International Speech Communication Association, INTERSPEECH, 2013, France, International Speech Communication Association, ISBN 9781629934433, 3756 p.
25-08-13_IS13_Program.pdf / 410.83 KB / No licence information
Keywords in Englishparalingustics, social signals, laughter detection, filler, support vector machines, Gaussian mixture models, cepstrum
ProjectThe Develpment of Digital Communicatios. Project leader: Lubacz Józef, , Phone: 22 234 65 31, start date 04-05-2012, planned end date 31-03-2013, end date 31-12-2013, IT/2012/statut, Completed
WEiTI Działalność statutowa
Languageen angielski
File
A.Janicki, Non-linguistic Vocalisation Recognition, Interspeech 2013.pdf 130.23 KB
Score (nominal)5
Score sourcejournalList
ScoreMinisterial score = 0.0, 02-02-2020, BookChapterSeriesAndMatConfByIndicator
Ministerial score (2013-2016) = 5.0, 02-02-2020, BookChapterSeriesAndMatConfByIndicator
Publication indicators WoS Citations = 3; Scopus Citations = 3; GS Citations = 11.0
Citation count*12 (2020-09-02)
Cite
Share Share

Get link to the record


* presented citation count is obtained through Internet information analysis and it is close to the number calculated by the Publish or Perish system.
Back
Confirmation
Are you sure?