APPLYING SOURCE-FILTER MODEL IN CHINESE SPEECH SYNTHESIS

This paper presents a novel algorithm for the signal modification component of concatenative text-to-speech systems. The algorithm described here is based on the LPC analysis/synthesis framework, and achieves prosodic modification by time-domain processing of the LPC residual(LF-4 model). This method is based on a source-filter model. The proposed method achieves high quality prosody modification, retains the characteristics of the donor speaker, allows for spectral manipulation, and yields compact acoustic inventories and improved voiced fricatives. Key Words: Speech synthesis Linear prediction Derivative glottal flow Prosody contro