Title :
Predication of prosodic data in Persian text-to-speech
Author :
Farrokhi, A. ; Ghaemmaghami, S.
Author_Institution :
Electr. Eng. Dept., Azad Univ., Tehran, Iran
Abstract :
A simplified four-layer recurrent neural network (RNN)
Keywords :
learning (artificial intelligence); multilayer perceptrons; speech synthesis; Persian text-to-speech systems; automatic segmentation; energy contour; four-layer recurrent neural network; natural sounding; onset time; pause duration; phonological states; pitch contours; prosodic data predication; prosodic information generator; prosodic parameters; root mean square error; speech synthesis; syllable length; syllable level; tang-sigmoid function; training function; vowel length; word level;
Journal_Title :
Electronics Letters
DOI :
10.1049/el:20031151
