Tones and break indices(ToBI) generation for Assamese sentences

摘要

The most important factor to build a synthesized discourse close to human speech, it is important that the content preparing segment delivers a suitable arrangement of phonemic units relating to an the text given as input. A detailed experimental research has been done throughout the project to syllabify Assamese words by using existing Phonetic Alphabet structures for Assamese Language and later comparing the results and there after extracting Assamese sentences from a mixed corpus of Assamese and English words. Speech is then recorded for the sentences and various features are extracted from the acoustic and text data in Assamese.