[Subject Prev][Subject Next][Thread Prev][Thread Next][Date Index][Thread Index]

[hts-users:04295] HTS-demo_NIT-SONG070 fails to generate intelligible speech


Hi all,

I trained HTS-demo_NIT-SONG070-F001 singing voice demo from the HTS website, with default configurations. The training was successful (no error nor warning appeared) but the output speech (gen/qst001/ver1/hts_engine/phrase01.wav) sounds weird. I've uploaded it here,
https://mega.co.nz/#!WUxXDRzZ!XHrevJ3MpVFS8KLJzYAJXqaCgzQN8No9rM4vRqj8XU8

Unlike the singing voice demo, HTS-demo_CMU-ARCTIC-SLT is able to synthesize natural and intelligible speech.
I'm running HTS-2.2 and HTK-3.4.1 on Ubuntu 14 (64bit).

Can anyone suggest possible reasons and how to get rid of this?

Thanks in advance and sincerely,
Kanru Hua