Methods of Delivering Text on Educational Platforms For Speech Synthesis

Abstract

The article provides a comprehensive examination of the orthoepic and prosodic features of speech synthesis in the Kazakh language. The primary goal of speech synthesis is to render text in a natural and intelligible spoken form. To achieve this, it is essential to explore orthoepic norms and prosodic characteristics from a theoretical perspective and analyze them in practical terms. During synthesis, it is necessary to systematically take into account such orthoepic phenomena as vowel harmony, the articulation of adjacent and merged consonants, as well as the reduction and voicing of sounds between words. In order to adapt orthoepic and prosodic patterns to synthesis software, models and patterns based on phonetic transcription are employed. Transcriptional annotations and models that accurately reflect the phonetic features of sounds contribute to the advancement of automatic speech synthesis. The article also addresses the issue of prosody, as it is a crucial factor in ensuring the naturalness and quality of synthesized speech. In this regard, maintaining the intonational features of speech directly affects the naturalness of the synthesized text. Segmenting texts into syntagmatic and phrasal groups and rendering them with attention to intonational components facilitates listener comprehension and enables a fuller understanding of the content. The article has practical significance, as the need for speech synthesis is becoming increasingly relevant in today's digital age. Speech synthesis is widely used in educational platforms, multimedia resources, programs for individuals with special needs, language learning systems, audiobooks, and virtual assistants.

Keywords: speech synthesis, prosody, orthoepy, phoneme, intonation.