Second European Conference on Speech Communication and Technology

Genova, Italy
September 24-26, 1991


Two Level Continuous Speech Recognition Using Demisyllable-Based HMM Word Spotting

Eduardo Lleida (1), Jose B. Marino (1), Climent Nadeu (1), Albert Oliveras (2)

(1) Dept. of Signal Theory and Communications, Universidad Politecnica de Catalunya, Barcelona, Spain
(2) Dept. of Automatics and Systems Engineering, Universidad Politecnica de Catalunya, Barcelona, Spain

This paper describes a two level Spanish Continuous Speech Recognition System based on Demisyllable HMM modelling, word-spotting and finite-state lexical and syntactic knowledge. The first level, the word level, is based on a spotting algorithm which takes as input the unknown utterance, the HMM of the reference demisyllable and the lexical knowledge in terms of a finite-state network. The output of the word level is a lattice of word hypothesis [1]. The second level, the phrase level, searches in a time-synchronous procedure the best sentence that end at each time instant. It takes as input the word lattice and the syntactic knowledge in terms of a finite-state network, giving as output the best legal sentence. The proposal two-level system was tested recognizing the integers from 0 to 1000 in a speaker independent approach. We get a word accuracy of 93,2% with a sentence accuracy of 84. 5%. Keywords: Speech Recognition, Hidden Markov Model, Fuzzy Training, Demisyllable, Word-spotting, Multiple Hypothesis, Finite State Networks.

Full Paper

Bibliographic reference.  Lleida, Eduardo / Marino, Jose B. / Nadeu, Climent / Oliveras, Albert (1991): "Two level continuous speech recognition using demisyllable-based HMM word spotting", In EUROSPEECH-1991, 1199-1202.