ttsunion / deep-expression Goto Github PK

Is pretrained model available? Also are generated samples available?

这个模型靠谱吗？想做中文tts

Hi currently your scripts re generate the original wav.
I did not find the way to do a text to speech based on a random phrase different from the original one. Maybe still not implemented?

in the train.py, you get predicted wav by "ypred = sess.run(yhat, feed_dict = {x:labels, y:wavs})". I can't understand tts's input is text and output is a wav, your project do a tts mission?

Can you please decribe what should be changed to use non Asian Language?

This would be amazing if it works!!!
But I am sorry, I think it is not really doing TTS. It requires the wave as input to the model which is not possible in prediction.

After running the training, I have tested the model prediction by filling y with zeros, it produces nothing. I looked at your reply issue #4, still not convinced. Could you please provide inference code to prove it?

I have heard the result sample based on only 4K steps training model. Unfortunately there are still some noises. I have not looked into the framework closely. But here are some suggestions from an open tacotron project about training data set. I think maybe you need to fetch better training samples such as THCHS30 and CVTE for this brilliant framework.