Voice files

There are over 700 voice sounds including Japanese and those from the previous work, although there are many different patterns of the same word. (There are more than 10 'Ah' alone.)
Moreover, There are many voices that could not be adopted, I would like to make something like a voice test function in the future.
In terms of operation, the voice algorithm was improved to create a voice cache so that the same voice is not continuously output as possible.