Publication Type:

Conference Paper

Source:

APSIPA ASC 2010 - Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, Biopolis, p.458-461 (2010)

URL:

http://www.scopus.com/inward/record.url?eid=2-s2.0-79958177295&partnerID=40&md5=fbfd0f9f2f4e4ef67f975489f1ade86a

Keywords:

Data processing, Feature extraction, Hidden Markov models, Labeled data, Multilingual acoustic models, Neural networks, Phone recognition, Probabilistic features, Speech recognition, Telephone circuits, Telephone sets, Telephone systems

Abstract:

In this paper, we study how performance of hidden Markov model - neural network (HMM-NN) phone recognizers can be enhanced using probabilistic features, without actually increasing the number of nodes in the neural network. This is necessary when the amount of labeled data available for training the models is small. We conduct two studies. One is to explore a multilingual probabilistic feature frontend. Another is to develop a multilingual acoustic model. We got an improvement of 2.87 and 4.75 per cent for Hindi and Tamil absolute phone recognition accuracy, and 3.03 and 7.02 per cent improvement for the multilingual phone recognition system for the respective languages.

Notes:

cited By (since 1996)0; Conference of org.apache.xalan.xsltc.dom.DOMAdapter@68d39971 ; Conference Date: org.apache.xalan.xsltc.dom.DOMAdapter@1bfeff7b Through org.apache.xalan.xsltc.dom.DOMAdapter@236f299; Conference Code:85090

Cite this Research Publication

C. P. Sa Kumar, Ambikairajah, Eb, Nosratighods, Mb, and Li, Hc, “Enhancing phone recognition accuracy using probabilistic features”, in APSIPA ASC 2010 - Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, Biopolis, 2010, pp. 458-461.