In this paper, it is our aim to define a set of objective acoustic criteria, based on subjective listeners assessment of talent voices, that can help to rate the voice font quality, bearing in mind the development of Text-to-Speech systems (hereafter TTS). The voice talent selection process is seldom discussed in scientific papers, since academic community is obviously more concerned with synthetic voice evaluation than with modal voice talent assessment. However, the choice of a voice that will be used in TTS systems is definitely of extreme importance. In this work, we describe the selection process of the European Portuguese voice talent, based on socio and linguistic criteria and supported by subjective and objective tests and measurements of modal voice quality. Relationship between subjective and objective parameters is presented and discussed. A subjective test comparing natural and resynthesized voices was also conducted and its results are discussed

--

voice font quality,voice talent,TTS,MLDC,signal processing,Text-to-Speech

• Ciências da Computação e da Informação - Ciências Naturais
• Engenharia Eletrotécnica, Eletrónica e Informática - Engenharia e Tecnologia
• Línguas e Literaturas - Humanidades