search

xcmyz / FastSpeech

The Implementation of FastSpeech based on pytorch.
MIT License
858 stars 213 forks source link

synthesized sentence is not complete #75

Closed zhoulinmin closed 4 years ago

zhoulinmin commented 4 years ago

i tried Chinese audio ,but the result is not very good, produced audio is only part of the sentence ,is this due to duration or attention , i used tacotron to produce the duration.

xcmyz commented 4 years ago

Check your duration target.