Automatic segmentation quality improvement for realization of unit selection speech synthesis

This paper reports progress in the process of improving automatic segmentation. To help achieve natural sounding speech synthesis not only construction of rich database is important but also precise alignment must be conducted. Process of automatic segmentation often causes errors that needs to be corrected. Manual correction doesnpsilat guarantee that the process will be carried out consistently. Script for finding outliers was constructed and only necessary manual correction was prepared. Then a praat script was realized which allowed to detect and remove most important errors in automatic segmentation. Zero crossing process was applied to each phoneme. Additionally a comparison in prototype speech synthesizer was performed. The results shows that quality of generated speech has significantly increased.