Fix normalize (#1404)

GPT_SoVITS/inference_webui.py

@@ -311,14 +311,14 @@ def get_phones_and_bert(text,language,version):
         if language == "zh":
             if re.search(r'[A-Za-z]', formattext):
                 formattext = re.sub(r'[a-z]', lambda x: x.group(0).upper(), formattext)
-                formattext = chinese.text_normalize(formattext)
+                formattext = chinese.mix_text_normalize(formattext)
                 return get_phones_and_bert(formattext,"zh",version)
             else:
                 phones, word2ph, norm_text = clean_text_inf(formattext, language, version)
                 bert = get_bert_feature(norm_text, word2ph).to(device)
         elif language == "yue" and re.search(r'[A-Za-z]', formattext):
                 formattext = re.sub(r'[a-z]', lambda x: x.group(0).upper(), formattext)
-                formattext = chinese.text_normalize(formattext)
+                formattext = chinese.mix_text_normalize(formattext)
                 return get_phones_and_bert(formattext,"yue",version)
         else:
             phones, word2ph, norm_text = clean_text_inf(formattext, language, version)

GPT_SoVITS/text/chinese.py

@@ -47,6 +47,19 @@ def replace_punctuation(text):
 
     replaced_text = pattern.sub(lambda x: rep_map[x.group()], text)
 
+    replaced_text = re.sub(
+        r"[^\u4e00-\u9fa5" + "".join(punctuation) + r"]+", "", replaced_text
+    )
+
+    return replaced_text
+
+
+def replace_punctuation_with_en(text):
+    text = text.replace("嗯", "恩").replace("呣", "母")
+    pattern = re.compile("|".join(re.escape(p) for p in rep_map.keys()))
+
+    replaced_text = pattern.sub(lambda x: rep_map[x.group()], text)
+
     replaced_text = re.sub(
         r"[^\u4e00-\u9fa5A-Za-z" + "".join(punctuation) + r"]+", "", replaced_text
     )
@@ -171,6 +184,20 @@ def text_normalize(text):
     return dest_text
 
 
+# 不排除英文的文本格式化
+def mix_text_normalize(text):
+    # https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/paddlespeech/t2s/frontend/zh_normalization
+    tx = TextNormalizer()
+    sentences = tx.normalize(text)
+    dest_text = ""
+    for sentence in sentences:
+        dest_text += replace_punctuation_with_en(sentence)
+
+    # 避免重复标点引起的参考泄露
+    dest_text = replace_consecutive_punctuation(dest_text)
+    return dest_text
+
+
 if __name__ == "__main__":
     text = "啊——但是《原神》是由,米哈\游自主，研发的一款全.新开放世界.冒险游戏"
     text = "呣呣呣～就是…大人的鼹鼠党吧？"

GPT_SoVITS/text/chinese2.py

@@ -60,6 +60,26 @@ def replace_punctuation(text):
     return replaced_text
 
 
+def replace_punctuation_with_en(text):
+    text = text.replace("嗯", "恩").replace("呣", "母")
+    pattern = re.compile("|".join(re.escape(p) for p in rep_map.keys()))
+
+    replaced_text = pattern.sub(lambda x: rep_map[x.group()], text)
+
+    replaced_text = re.sub(
+        r"[^\u4e00-\u9fa5A-Za-z" + "".join(punctuation) + r"]+", "", replaced_text
+    )
+
+    return replaced_text
+
+
+def replace_consecutive_punctuation(text):
+    punctuations = ''.join(re.escape(p) for p in punctuation)
+    pattern = f'([{punctuations}])([{punctuations}])+'
+    result = re.sub(pattern, r'\1', text)
+    return result
+
+
 def g2p(text):
     pattern = r"(?<=[{0}])\s*".format("".join(punctuation))
     sentences = [i for i in re.split(pattern, text) if i.strip() != ""]
@@ -171,6 +191,23 @@ def text_normalize(text):
     dest_text = ""
     for sentence in sentences:
         dest_text += replace_punctuation(sentence)
+
+    # 避免重复标点引起的参考泄露
+    dest_text = replace_consecutive_punctuation(dest_text)
+    return dest_text
+
+
+# 不排除英文的文本格式化
+def mix_text_normalize(text):
+    # https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/paddlespeech/t2s/frontend/zh_normalization
+    tx = TextNormalizer()
+    sentences = tx.normalize(text)
+    dest_text = ""
+    for sentence in sentences:
+        dest_text += replace_punctuation_with_en(sentence)
+
+    # 避免重复标点引起的参考泄露
+    dest_text = replace_consecutive_punctuation(dest_text)
     return dest_text