PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Analysing recognition errors in unlimited-vocabulary speech recognition
Teemu Hirsimäki and Mikko Kurimo
In: 2009 Annual Conference of the North american Chapter of the association for Computational Linguistics, NAACL 2009(2009).


We analyze the recognition errors made by a morph-based continuous speech recognition system, which practically allows an unlim- ited vocabulary. Examining the role of the acoustic and language models in erroneous regions shows how speaker adaptive training (SAT) and discriminative training with mini- mum phone frame error (MPFE) criterion de- crease errors in different error classes. An- alyzing the errors with respect to word fre- quencies and manually classied error types reveals the most potential areas for improving the system.

PDF - PASCAL Members only - Requires Adobe Acrobat Reader or other PDF viewer.
EPrint Type:Conference or Workshop Item (Paper)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Natural Language Processing
ID Code:6056
Deposited By:Mikko Kurimo
Deposited On:08 March 2010