FastSpeech2 icon indicating copy to clipboard operation
FastSpeech2 copied to clipboard

inference problom about energy and pitch

Open fenling opened this issue 4 years ago • 1 comments

hi ming024 i have some questions about inference。 1)energy and pitch change ,but the generat wav is similiar. i set pitch 0.7、1.0、5, the wav sounds similiar.         "--pitch_control",         type=float,         default=5.0, i set it error? or the pitch change is small. if i want big pitch change, how should i do. 2) i download the AISHELL3_600000.pth, how long did you train this model. how many are the batchsize and the step?

fenling avatar Jun 09 '21 10:06 fenling

hi ming024 i have some questions about inference。 1)energy and pitch change ,but the generat wav is similiar. i set pitch 0.7、1.0、5, the wav sounds similiar.         "--pitch_control",         type=float,         default=5.0, i set it error? or the pitch change is small. if i want big pitch change, how should i do. 2) i download the AISHELL3_600000.pth, how long did you train this model. how many are the batchsize and the step?

Hi @fenling Did you solve this problem? I encountered the same issue.

zheng-xing avatar Mar 03 '22 09:03 zheng-xing