Synthesis of speech from Bangla text

Abstract

BangIa text to speech synthesizer can generate speech sounds from arbitrary text input. BangIa text to speech synthesizer is developed using the method that stores speech sounds of parts of words. The parts of words i.e. basic sounds units are recorded using sound card, microphone and sound recorder utility software. The pronunciation rules and the determination of basic sound units of BangIa words are studied extensively as part of the current research which is the field of BangIa phoneme research. At first BangIa text to speech synthesizer accepts the character from keyboard and identifies the sound unit and plays that. This task is repeated until the word is completed. Then the text is normalized by defining the wave files from the adjacent keys as per pronunciation rule and searched. When identification of the sound units for a particular BangIa word is completed, all the sound units are merged together into a speech file to form the complete speech signal for that Bangia word and finally the merged wave file is played. Note that BangIa text to speech synthesizer is interfaced with the applicationlword processing software so that at first the message (generated by pressing key) is retrieved from application's message queue to BangIa text to speech synthesizer.