Fix bug in #1660 and #1667 (#1670)

* fix #1660 #1667

* add japanese cmudict-katakana dict

GPT_SoVITS/text/japanese.py

@@ -1,9 +1,9 @@
 # modified from https://github.com/CjangCjengh/vits/blob/main/text/japanese.py
 import re
-
-import pyopenjtalk
 import os
 import hashlib
+try:
+    import pyopenjtalk
     current_file_path = os.path.dirname(__file__)
     def get_hash(fp: str) -> str:
         hash_md5 = hashlib.md5()
@@ -24,6 +24,11 @@ if os.path.exists(USERDIC_CSV_PATH):
 
     if os.path.exists(USERDIC_BIN_PATH):
         pyopenjtalk.update_global_jtalk_with_user_dict(USERDIC_BIN_PATH)   
+except Exception as e:
+    # print(e)
+    import pyopenjtalk
+    # failed to load user dictionary, ignore.
+    pass
 
 
 from text.symbols import punctuation
@@ -80,10 +85,6 @@ def post_replace_ph(ph):
 
     if ph in rep_map.keys():
         ph = rep_map[ph]
-    # if ph in symbols:
-    #     return ph
-    # if ph not in symbols:
-    #     ph = "UNK"
     return ph
 
 
@@ -103,6 +104,8 @@ def symbols_to_japanese(text):
 def preprocess_jap(text, with_prosody=False):
     """Reference https://r9y9.github.io/ttslearn/latest/notebooks/ch10_Recipe-Tacotron.html"""
     text = symbols_to_japanese(text)
+    # English words to lower case, should have no influence on japanese words.
+    text = text.lower()
     sentences = re.split(_japanese_marks, text)
     marks = re.findall(_japanese_marks, text)
     text = []
@@ -219,5 +222,5 @@ def g2p(norm_text, with_prosody=True):
 
 
 if __name__ == "__main__":
-    phones = g2p("こんにちは, hello, AKITOです,よろしくお願いしますね！")
+    phones = g2p("Hello.こんにちは！今日もNiCe天気ですね！tokyotowerに行きましょう！")
     print(phones)