Multilingual phone models for vocabulary-independent speech recognition tasks

This paper presents three different methods for developing multilingual pho ne models for flexible speech recognition tasks. The main goal of our inves tigations is to find multilingual speech units that work equally well in ma ny languages. With such a universal set it is possible to build speech reco gnition systems for a variety of languages. One advantage of this approach is that acoustic-phonetic parameters in a HMM-based speech recognition syst em can then be shared. The multilingual approach starts with the phone sets of six languages, a total of 232 language-dependent and context-independen t phone models. Then, we develop three different methods to map the languag e-dependent models to a multilingual phone set. The first method is a direc t mapping to the phone set of the International Phonetic Association (IPA). In the second approach we apply an automatic clustering algorithm for the phone models. The third method exploits the similarities of single mixture components of the language-dependent models. Like the first method the lang uage-specific models are mapped to the IPA inventory. In the second step an agglomerative clustering is performed on the density level to find regions of similarity between the phone models of different languages. The experim ents carried out with the SpeechDat(M) database, show that the third method yields almost the same recognition rate as language-dependent models. Howe ver, using this method we achieve a huge reduction of the number of densiti es in the multilingual system. (C) 2001 Elsevier Science B.V. All rights re served.