oken in most of the major cities, the need for developing the multilingual speech recognition systems is becoming increasingly important. Some of the studies claim that more than half of the world's population is multilingual [1]. This claim would be more appropriate to a multilingual country like India, which is a home for large number of languages. India has 122 major languages [2] among them 22 are constitutionally recognised [3]. Since most of the Indian languages belong to one of the four language families - Indo-Aryan, Dravidian, Austro-Asian, and Sino-Tibetan [3] - and the languages within a language family have large linguistic similarities in their grammatical structures and vocabulary patterns [4], the development of multilingual speech recognition systems is very appropriate in the context of Indian languages. Hence, it is essential to investigate on the development of multilingual speech recognition systems using Indian language