search

nii-yamagishilab / self-attention-tacotron

An implementation of "Investigation of enhanced Tacotron text-to-speech synthesis systems with self-attention for pitch accent language" https://arxiv.org/abs/1810.11960
BSD 3-Clause "New" or "Revised" License
113 stars 32 forks source link

Do you guys have any samples to upload #5

Closed cliuxinxin closed 5 years ago

cliuxinxin commented 5 years ago

That will make it clear.

TanUkkii007 commented 5 years ago

@cliuxinxin We can share some samples from LJSpeech. The Japanese corpus we used is not redistributable so we cannot provide Japanese samples in public.

cliuxinxin commented 5 years ago

I think it's a good idea. We all have LJspeech ：)

TanUkkii007 commented 5 years ago

@cliuxinxin Sorry for the late response. I created audio samples from LJSpeech. Audio quality of the samples is not good so I keep trying to improve it. In addition I will addd samples from other datasets in the future. https://nii-yamagishilab.github.io/self-attention-tacotron-audio-samples/index.html

cliuxinxin commented 5 years ago

@TanUkkii007 Thank you for your reply. I have heard the samples sounds good. Any chance to implement the gst? I think add gst maybe let the sounds more nature.If I got some time can try your code on my own dataset , and give you some feedback. After Chinese spring festival. sank you for your work.

TanUkkii007 commented 5 years ago

@cliuxinxin gst stands for global style tokens, right? I may implement it for a project but that project is not active so please do not expect too much. I look forward to your feedback on your dataset.