Text To Speech Conversion - Tamil Language

Abstract

A text-to-speech conversion engine was developed for Tamil language that could successfully convert any arbitrary Tamil text into spoken utterance. The back-end of the engine was a sound database created with 343 phonemes which were chosen as the basic units of Tamil language. The text-to-speech engine splits the given text into the basic units of Tamil, extracts the correct phonemes from the database and concatenates the basic unit waveforms in the correct sequence to produce the required sound waveform. The front-end of the engine is a GUI interface that allows users to input Tamil text in letters or even numbers. The text can be stored to be used at a later time. The Tamil font “TSC_Avarangal” was selected as the font to develop the user interface. The sound editing software “Sound Forge” was used when preparing the phonemes with correct pitch. This work shows that for languages such as Tamil or Sinhala textto- speech engines can be successfully developed with the available resources.