This paper describes a speech synthesizer which combines parametric speech synthesis with speech synthesis in the time domain. In the proposed system, the alternate or simultaneous use of both synthesis methods is controlled by a phoneme-dependent algorithm. Vowels and nasals are produced by the parametric system part - a formant-based synthesizer. Voiceless consonants are generated in the time domain. To synthesize voiced fricatives, voiced plosives and voiced/voiceless-transitions, both system components work simultaneously.
Keywords: speech synthesis, formant synthesizer, hybrid system