VowSynth A Synthesizer of Vowel Sounds Based on Additive Synthesis
E. S. Guerra · 1998
This article presents the design and implementation of a non real-time voice synthesizer based on the analysis of a soprano voice singing the five spanish vowels. The analysis is based on the SMS technique and the synthesis uses additive synthesis. 1 Introduction The synthesis of the singing voice has been investigated from two different fields, speech and music, and following mainly two different models: physical models, based on sound productions mechanisms, and spectral models, based on sound perception mechanisms. (For a general overview of singing voice synthesis methods see [1]). One of the first methods to approach singing voice synthesis was LPC, which can be considered a mixture of both types of models. Other early approaches were mainly based on spectral models, like the vocoder, FM, CHANT, or MUSSE. SPASM can be included in physical models where the human vocal tract is modeled as a digital waveguide. SMS (Spectral Modeling Synthesis) [2] is also a spectral model, and the b...