Employing MaryTTS to Synthesize the Tamil Language and Narrate Children’s Stories

Abstract

Over five million individuals of Sri Lanka, or 15% of the entire population, speak Tamil as their native language, where the particular language is considered to be the country’s second official language. When given a task to speak or read a language, kindergarteners may struggle to read and learn a language on their own. Especially in their native language. For example, Tamil. Play schools taught in English are hosted across the island, thus this could be one of the reasons. Storytelling is a formative experience for children. Early on, stories can influence a child's growth. Reading helps people gain knowledge and language. Reading is said to help kids learn languages. Due to the compelling nature of storytelling, it can be utilized in both Early Childhood Education and Primary Education. This is because both of these phases require knowledge transmission. Storytelling is an essential instructional tool, as well as a significant means of preserving culture and transmitting essential values to future generations. Currently, the literary plot must be updated in order to suit the expectations of a new audience, which exists in a culture where knowledge and information are abundant. Tamil is a difficult language to master when compared to other spoken languages due to its complex grammatical structure. In the context of the Tamil language, natural language processing (NLP) technologies are still in their early phases of development. Many other languages, spoken worldwide, have efficient computer systems that can operate in their native language, despite today's more complex and language-independent Text-to-Speech frameworks. In this situation, Tamil text-to-speech could be used to communicate with the children. During the first few years of a child's life, the child's mental development should be a top priority. There are only a few computer systems that can work with the Tamil language. Even though there are a lot of apps for telling stories, most of them are mostly made for the English language. The objective of the research is consequently to create an application for children that uses text-to-speech (TTS) technology to convey stories. Therefore, the research focuses on the significance of text-to-speech technology in supporting kids with the narration of a story. Once the storyteller application was implemented, it was compared to human speech. The results of the paired samples t-tests conducted on the TTS-generated speech samples and the human-produced speech samples revealed that there is a statistical significance exists between the two types of samples for every metric (speech quality), which means that there is no acceptable amount of difference between the TTS produced story (speech) and the human produced story (speech). The findings of this research indicate that modern TTS systems appear ready to proceed to next stages of evaluation once the speech quality reaches the necessary level.