Third International Conference on Spoken Language Processing (ICSLP 94)

Yokohama, Japan
September 18-22, 1994

Hidden Markov Models and Selectively Trained Neural Networks for Connected Confusable Word Recognition

Jean-Frangois Mari (l), Dominique Fohr (l), Yolande Anglade (l), Jean-Claude Junqua (2)

(1) CRIN-CNRS & INRIA Lorraine, Vandoeuvre-les-Nancy, France
(2) Speech Technology Laboratory, Div. of Panasonic, Santa Barbara, CA, USA

This paper presents a new method for connected-word recognition with confusable vocabularies, such as connected letters. The recognition process is performed in two steps. First, a second-order HMM provides N-best word strings. Then, the strings of confusable letters are discriminated by a procedure based on acoustic knowledge and artificial neural networks (ANN). This method has been tested on an American-English database containing spelled names collected through the telephone network. The results obtained with the first HMM pass and the improvements made with the ANN are presented and discussed. When a 3,300 name dictionary and a retrieval procedure based on a DTW alignment algorithm were used, 96% recognition accuracv was obtained.

Full Paper

Bibliographic reference.  Mari, Jean-Frangois / Fohr, Dominique / Anglade, Yolande / Junqua, Jean-Claude (1994): "Hidden Markov models and selectively trained neural networks for connected confusable word recognition", In ICSLP-1994, 1519-1522.