Spectral Smoothening Based Waveform Concatenation Technique for Speech Quality Enhancement in Text-to-Speech Systems

This work presents a spectral smoothening based concatenation technique for enhancing the quality of speech produced by Text-to-Speech systems. As, the hard waveform concatenation process may cause audible glitches at the segment boundaries affecting the overall quality of the produced speech, an optimal coupling based spectral smoothening approach is adopted to smoothen the spectral envelop of the produced speech for enhancing its quality. A number of experiments were performed to analyze the performance of the proposed technique for which different speech quality evaluation parameters are considered and the results are compared with the other con-catenative techniques. The results obtained in all the experiments performed shows the effectiveness of the proposed text-to-speech conversion technique in producing high-quality results.