Change search
ReferencesLink to record
Permanent link

Direct link
An Analysis of Shallow and Deep Representations of Speech Based on Unsupervised Classification of Isolated Words
KTH, School of Computer Science and Communication (CSC), Speech, Music and Hearing, TMH, Speech Communication and Technology.ORCID iD: 0000-0002-3323-5311
2015 (English)In: Proceedings of Nonlinear Speech Processing, Springer, 2015Conference paper (Refereed)
Abstract [en]

We analyse the properties of shallow and deep representa-tions of speech. Mel frequency cepstral coefficients (MFCC) are compared to representations learned by a four layer Deep Belief Network (DBN) in terms of discriminative power and invariance to irrelevant factors such as speaker identity or gender. To avoid the influence of supervised statistical modelling, an unsupervised isolated word classification task is used for the comparison. The deep representations are also obtained with unsupervised training (no back-propagation pass is performed). The results show that DBN features provide a more concise clustering and higher match between clusters and word categories in terms of adjusted Rand score. Some of the confusions present with the MFCC features are, however, retained even with the DBN features.

Place, publisher, year, edition, pages
Springer, 2015.
National Category
Computer Science Language Technology (Computational Linguistics)
Identifiers
URN: urn:nbn:se:kth:diva-180414DOI: 10.1007/978-3-319-28109-4_15ScopusID: 2-s2.0-84955471729OAI: oai:DiVA.org:kth-180414DiVA: diva2:893723
Conference
Nonlinear Speech Processing
Note

QC 20160615

Available from: 2016-01-13 Created: 2016-01-13 Last updated: 2016-06-15Bibliographically approved

Open Access in DiVA

No full text

Other links

Publisher's full textScopushttps://sites.google.com/site/nolisp2015/home

Search in DiVA

By author/editor
Salvi, Giampiero
By organisation
Speech Communication and Technology
Computer ScienceLanguage Technology (Computational Linguistics)

Search outside of DiVA

GoogleGoogle Scholar
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Altmetric score

Total: 11 hits
ReferencesLink to record
Permanent link

Direct link