International Journal of Advanced Research in Computer Science and Software Engineering

Abstract

The main purpose of the study was to develop a speech recognition system for isolated digits of English language using HTK. Speech, in addition to being a tool of communication, is also a symbol of identity and authorization. Two different corpora were collected of audio recordings of isolated digits of English language speakers, in which speakers read numeric digits. Both of the collected corpora contained the training data and the other testing data. One corpus is self recorded signals and other is standard CUAVE dataset (50 speakers, each uttered 10 words). The system has been implemented using the HMM toolkit i.e. HTK by training HMMs of the words making the vocabulary on the training data. Different HMMs for individual digits have been initialized and trained to have well modelled structure. The trained system was tested on training data as well as test data and results revealed that 95% of the data was correctly recognized. The developed system can be used by developers and researchers interested in speech recognition for English language not only for isolated digits but also for other words of English language. The findings of the study can be generalized to cater for large vocabularies and for continuous speech recognition.