Georgian test to speech

Author: David Abulashvili
Keywords: text to speech, neural networks, generative models, dataset
Annotation:

In this research we provide one of the most interesting problems: text to speech. Relevance of this task and importance for speakers of Georgian language is shown. Various potential uses and implications of the product are provided. Different ways to solve this task are discussed, including classical programing approach, as well as artificial intelligence technologies. The processes of data collection, processing, and analysis as well as the results of the experiment conducted are presented in this paper. The steps taken for training the neural networks and challenges faced during the process are also discussed. The results are summarized, future goals are set and plan for achieving goals is provided.

Lecture files:

ქართული ტექსტის ხმად გარდაქმნა [ka]