search

ttsunion / Deep-Expression

An Attention Based Open-Source End to End Speech Synthesis Framework, No CNN, No RNN, No MFCC!!!
85 stars 27 forks source link

non Asian languages #3

Open alain2208 opened 6 years ago

alain2208 commented 6 years ago

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

alain2208 commented 6 years ago

In fact I found it myself. replace in parameters.py vocab_size = 28 i think 26 + % + space is that right? in text.py replace string by latin alphabet plus space plus%

FonzieTree commented 6 years ago

@alain2208 Yeah, you are right, have a try, looking forward to know your result.

FonzieTree commented 6 years ago

@alain2208 , please download the latest version, noise problem almost disappear in this demo.

alain2208 commented 6 years ago

i made a personal record and increase the level got very good quality. As your test are in Chinese I think (sorry if it's not chinese) it's not very easy for non chinese to work. How is it suppose to work It loads 4 files but only try to reproduce the first one that's right? Nice job! very impressed to hear my voice back by a computer. even after 4000 run already got very good result.

FonzieTree commented 6 years ago

Thank you for your feedback.

edwinlimlx commented 6 years ago

@alain2208 as mentioned the following change for EN language vocab_size = 28 text.py replace string by latin alphabet plus space plus%

does that mean your samples/labels are just abc...? how can we make it speak via words?