WaveGRU-Text-To-Speech / extract_model.py
NTT123
new model
e128a73
raw
history blame
186 Bytes
import pickle
dic = pickle.load(open("./wavegru_vocoder_1024_v3_1330000.ckpt", "rb"))
del dic["optim_state_dict"]
pickle.dump(dic, open("./wavegru_vocoder_1024_v3_1330000.ckpt", "wb"))