search

cnlinxi / style-token_tacotron2

style token with tacotron2
MIT License
61 stars 16 forks source link

input the same text but output different #7

Closed kanatazhan closed 5 years ago

kanatazhan commented 5 years ago

Thanks for you work, and when I synthesize audio with the same text, I found that the outputs sound a little different, so I checked the inference from session.run and found that the same input leader to different output. Do you know why that happen?

cnlinxi commented 5 years ago

I donot know what do you mean. You should run python synthesize.py --text_list <your corpus file> --name <your model name>. If sentence in is different, the synthesized voice is different.