A prosodic Text-to-Speech system for Yorùbá language

Abstract

We reviewed prosody realization in Text-to-Speech (TTS) applications. Standard Yoruba (SY) was considered alongside. Our review showed that the language is under researched in the area of prosody and speech synthesis generally. Some technologies that produced good prosody in certain syllable based languages were identified. These include the use of polysyllabic units, diphones and Hidden Markov's Model (HMM). We recommended a fusion of these technologies to realize the Yoruba TTS of high prosody. A mean opinion score (MOS) is also proposed to ascertain naturalness of the proposed TTS. This report presents among others, the review of prosody techniques in TTS, challenges of a Yoruba TTS and the proposed solution.

BibleTTS: a large, high-fidelity, multilingual, and uniquely African speech corpus

“…TWB Gamayun kits [17] include a 6-hour single speaker high-quality Swahili speech corpus optimized for TTS training. Earlier Yorùbá TTS efforts typically used bespoke private data [18,19,20,21,22]. For isiXhosa, Sesotho, Setswana and Afrikaans, multi-speaker corpora of approximately 2 hours each have been developed for TTS [23,24].…”