Data Processing for Optimizing Naturalness of Vietnamese Text-to-speech System

Abstract End-to-end text-to-speech (TTS) systems has proved its great success\nin the presence of a large amount of high-quality training data recorded in\nanechoic room with high-quality microphone. Another approach is to use\navailable source of found data like radio broadcast news. We aim to optimize\nthe naturalness of TTS system on the found data using a novel data processing\nmethod. The data processing method includes 1) utterance selection and 2)\nprosodic punctuation insertion to prepare training data which can optimize the\nnaturalness of TTS systems. We showed that using the processing data method, an\nend-to-end TTS achieved a mean opinion score (MOS) of 4.1 compared to 4.3 of\nnatural speech. We showed that the punctuation insertion contributed the most\nto the result. To facilitate the research and development of TTS systems, we\ndistributed the processed data of one speaker at\nhttps://forms.gle/6Hk5YkqgDxAaC2BU6.\n