You are on page 1of 23

Employing MaryTTS to

Synthesize Tamil Language and


Narrate Children Stories
Name: M.I. Fathima Nihla
Student ID: MS21900822
Program of Study: Master of Science in Information Technology
Supervisor: Prof. Koliya Pulasinghe

Sri Lanka Institute of Information Technology


2022
INTRODUCTI
ON
Introduction
• Tamil is Sri Lanka's second official language, spoken by around five million
people on the island, or about 15% of the total population.
• Children find it challenging to read and grasp a language on their own at the
kindergarten level. Particularly, their native language.
• One of the likely causes for this could be the English-language play schools
that are held all around the island.
• As a child, one of the most important things that everyone hears is a story.
Early on, stories play an important role in enhancing a child's learning.
• But Tamil is a difficult language to learn when compared to other languages
because of its complex grammar.
• In the context of the Tamil language, natural language processing (NLP)
technologies are still in their infancy.
• Implementing a Tamil text-to-speech technology for communication with kids
is a good alternative in such a case.
RESEARCH AIM &
OBJECTIVES
Research Aim & Objectives

• The aim of the present study is to create a Tamil TTS


system based on MaryTTS that can be used to narrate
children’s stories. As a result, the following study
objectives were devised in order to achieve the stated
goal.
A. To employ and adopt MaryTTS to synthesize Tamil
language.
B. To implement a Tamil Text-to-Speech system for
narrating stories to children.
C. To design and create a story-telling app that
includes Text-to-Speech technology to assist
children with reading Tamil stories.
RESEARCH
QUESTIONS
Research Questions

1. RQ1 - How can a basic set of natural language


processing (NLP) modules for the Tamil language be
developed?
2. RQ2 – How can Tamil language be configured on
MaryTTS?
3. RQ3 - How can a new Tamil synthesis voice using
MaryTTS be created?
4. RQ4 - What are the required resources and state-of-
the-art techniques for the speech synthesis?
PROGRE
SS
• Defining the Research Contribution
• Implementation of a Tamil TTS System
Research Contribution
• This part of the study makes two main
contributions.
• The first is proposing an algorithm
(methodology) on how MaryTTS can be
adopted to synthesize Tamil language
through a clear research and review.
• The second is an in-depth
documentation of the approach to be
followed in synthesizing a voice for an
under resourced language.
MARYTTS FRAMEWORK - METHODOLOGY
Proposed Algorithm

Algorithm:
1. First select the Linux as the environment
platform
2. Install the prerequisites
3. Configure MaryTTS
4. Configure Tamil language
5. Build the voice
Text-to-Speech

• The text-to-speech (TTS) is the process of converting words into a vocal


audio form.
• The program, tool, or software takes an input text from the user, and using
methods of natural language processing understands the linguistics of the
language being used and performs logical inference on the text.
• This processed text is passed into the next block where digital signal
processing is performed on the processed text.
• Using many algorithms and transformations this processed text is finally
converted into a speech format.
• This entire process involves the synthesizing of speech.
Block Diagram
• Installation of gTTS module
• Enter the Tamil story
How it • Text-to-speech translation
works? • Save the converted speech
• Run the file (play or read the story to
kid)
Tamil Storyteller
THESIS WRITING
TO-DO
• Test and Evaluation of the Research
• It is expected to conduct a user experiment test
on the implementation done.

• A questionnaire will be prepared and will


be presented to users.

• Complete the Thesis


• Complete the Research Paper
CONCLUSION
THANK YOU!

You might also like