An Automatic Spontaneous Live Speech Recognition System for Punjabi Language Corpus

In spontaneous Punjabi speech model, the speech is basically non-planed and non designed, there are generally depicted by repetitions, preservation, wrong start, half-spoken words and non-planned words, silence gap etc. In a system of Punjabi speech detection including vocabulary, the identification needs the evaluation among the audio signal of the utterance and the variety of utterances of the vocabulary. The primary purpose of the research paper is to build the automatic spontaneous speech recognition system for the Punjabi language. The Punjabi speech system is trained and tested by the voice of different male and female speakers and GUI for spontaneous Punjabi live speech system also has been developed. The performance analysis of the speech system is measured by using live speech method. The recognition accuracy for Punjabi speech sentences is 91.17% and 85.38% for different Punjabi words. The sphinx framework and java programming language are used to develop an automatic spontaneous Punjabi live speech model.