Hi, Cenk Demiroglu wrote (2008/09/26 4:43):
I have been trying to generate 8khz speech using alpha=0.32 and the 25 MCEP coefficients. Output speech sounds very muffled. The quality is not really acceptable compared to 16khz speech.I knowthat there should be some degradation when we go from 16 to 8 khz sampling rate but the difference seems to be huge.Am I doing something wrong? Are there other changes I should makebesides changing the alpha parameter? Is this a known issue with hts?
I think you should reduce the order of mel-cepstral analysis. 24-order mel-cepstral analysis sounds too high for 8kHz sampling. Regards, Heiga ZEN (Byung Ha CHUN) -- ------------------------------------------------ Heiga ZEN (in Japanese pronunciation) Byung Ha CHUN (in Korean pronunciation) ------------------------------------------------