search

mmorise / World

A high-quality speech analysis, manipulation and synthesis system
http://www.kisc.meiji.ac.jp/~mmorise/world/english
Other
1.18k stars 253 forks source link

Get bad wav after analysis and synthesis back #107

Closed Yablon closed 3 years ago

Yablon commented 3 years ago

I get bad wav after the following operations:

./build/analysis test.wav test.f0 test.sp test.ap
./build/synthesis test.f0 test.sp test.ap debug.wav

Do you have any idea on how to fix this ?

The original test wav and debug wav are as follows wav.zip

Thank you !

mmorise commented 3 years ago

F0 estimation error is the main cause of degradation. The sound quality would improve by using Harvest() instead of Dio(). Harvest() generally provides a more reliable F0 compared with Dio().

Yablon commented 3 years ago

@mmorise Thank you for your kind reply. How about f0 estimated by REAPER or Praat ? Is that as reliable as Harvest() ?

mmorise commented 3 years ago

REAPER and Praat would work well when the noise in speech is small. The sound quality of synthesized speech generally depends on the voiced/unvoiced estimation in the F0 estimation process. To obtain good sound quality, Harvest is designed to reduce the error that the voiced section is wrongly identified as the unvoiced section.

Yablon commented 3 years ago

@mmorise Thank you !