Design Issues in Automatic Grapheme-to-Phoneme Conversion for Standard Yorùbá

Grapheme-to-Phoneme (G2P) conversion is an important problem in Human Language Processing development, particularly Textto-Speech (TTS). Its primary goal is to accurately compute the pronunciation of words in the input texts. This work examines design issues with respect to components of the automatic G2P for standard Yorùbá (SY). The automatic process includes: (i) Tokenisation of Input, (ii) Identification of nasal characters, (iii) Syllabification, and (iv) Conversion of Graphemes to Phonemes. The structure of the Yoruba text is described and a text corpus design for standard Yorùbá TTS is presented. The analysis of the data was done using Zipf’s law. The outcome of this work provided adequate requirement for the design.