Correction for text with punctuation and dash

NVIDIA / mellotron

Mellotron: a multispeaker voice synthesis model based on Tacotron 2 GST that can make a voice emote and sing without emotive or singing training data

BSD 3-Clause "New" or "Revised" License

854 stars 187 forks source link

Correction for text with punctuation and dash #33

Open hyunjoolee opened 4 years ago

hyunjoolee commented 4 years ago

I have found that if there are punctuation and dash characters in the text, they are not converted to clean text in text/init.py get_arpabet().

For examples, words like "recommendations.", "fbi," and "policy-making" are not searchable in the cmu_dict. I think these will reduce model performance.

So I suggest some code as attached.