Program for Speech Synthesis

Development of methods for synthesizing human-like voice from text has been conducted for many years, but still remains relevant, because the solutions and technologies are constantly evolving. In order to create a synthesized speech, as close as possible to the human, in the development of speech synthesis algorithms, it is necessary to consider the specific structures and rules of the language for which it is developed, namely, the rules of transcription, the extraction of the minimum semantic units of the language (phonemes) and a huge amount of exceptions. The purpose of this study is to create a program that will allow to synthesize human-like speech on an unknown input text in Russian. To do this, it is necessary to study the basic methods of speech synthesis and summarize the knowledge to determine the most appropriate approach. The paper presents the existing methods of speech synthesis and describes their advantages and disadvantages, as well as defined and described in detail the chosen approach. The process of speech synthesis from the creation of a database containing the voice acting of Russian phonemes to the algorithms for obtaining audio files containing the synthesized speech from the text received at the input of the program is fully described. The result of the work is a program that allows you to get a synthesized speech in Russian on an unknown text. The program includes a web application to demonstrate the quality of the voice synthesizer and an open API that allows you to integrate the speech synthesis service on an incoming text request into almost any application.