Two Synthesis Methods Based on Cepstral Parameterization

A. Madlova, J. Pribil
2002 Radioengineering  
The paper deals with two implementations of the speech synthesis based on the cepstral representation of the human vocal tract model. Both the approaches to speech modeling are evaluated in the frequency domain. The paper also compares computational complexity of these two methods.
doaj:de250665e29e4b4a8487d99481df432a fatcat:5ujqxgvpujhd5o4nk7mygwm37u