Vowel onset point based waveform concatenation technique for intelligible speech synthesis

This paper describes an intelligible speech synthesis technique for Odia language. The model uses a vowel onset point identification technique to determine the fraction durations and is being integrated with the rule-based concatenative technique to produce speech out of the input text. To analyze the performance of the proposed technique, different types of words in the Odia language are considered. The results of all the experiments performed shows the effectiveness of the technique in producing intelligible speech in the considered language compared to the existing techniques.