Design An Optimization-Based Convolutional Neural System for Converting Text into Speech

Mukta Sandhu
Page No. : 819-835

ABSTRACT

The most challenging task in speech synthesis is background noise, very time-consuming, high cost, and power-consuming. To overcome these issues design a Butterfly-based Convolutional Neural System (BbCNS). Initially, the input text of certain users was collected and trained into the system and the preprocessing is utilized for removing the errors present in the dataset and preparing the text data for a specific context. Additionally, data normalization is employed to transfer the text into canonical and consistent form. Additionally, linguistic analysis is used to understand the content of the text and to identify the constituent morphemes of each word. Furthermore, Prosodic prominence prediction can be predicted from written language. Finally, the waveform is generated for converting text into speech. At last, the gained outcomes of the designed model are validated with other prevailing models with respect to accuracy, sensitivity, specificity, precision, and computation time.


FULL TEXT