Sync from GitHub repo

This Space is synced from the GitHub repo: https://github.com/SWivid/F5-TTS. Please submit contributions to the Space there

src/f5_tts/infer/utils_infer.py

@@ -346,7 +346,7 @@ def preprocess_ref_audio_text(ref_audio_orig, ref_text, clip_short=True, show_in
         else:
             ref_text += ". "
 
-    print("ref_text  ", ref_text)
+    print("\nref_text  ", ref_text)
 
     return ref_audio, ref_text
 
@@ -378,6 +378,7 @@ def infer_process(
     gen_text_batches = chunk_text(gen_text, max_chars=max_chars)
     for i, gen_text in enumerate(gen_text_batches):
         print(f"gen_text {i}", gen_text)
+    print("\n")
 
     show_info(f"Generating audio in {len(gen_text_batches)} batches...")
     return infer_batch_process(

src/f5_tts/model/utils.py

@@ -133,16 +133,23 @@ def get_tokenizer(dataset_name, tokenizer: str = "pinyin"):
 
 # convert char to pinyin
 
+jieba.initialize()
+print("Word segmentation module jieba initialized.\n")
+
 
 def convert_char_to_pinyin(text_list, polyphone=True):
     final_text_list = []
-    god_knows_why_en_testset_contains_zh_quote = str.maketrans(
-        {"“": '"', "”": '"', "‘": "'", "’": "'"}
-    )  # in case librispeech (orig no-pc) test-clean
-    custom_trans = str.maketrans({";": ","})  # add custom trans here, to address oov
+    custom_trans = str.maketrans(
+        {";": ",", "“": '"', "”": '"', "‘": "'", "’": "'"}
+    )  # add custom trans here, to address oov
+
+    def is_chinese(c):
+        return (
+            "\u3100" <= c <= "\u9fff"  # common chinese characters
+        )
+
     for text in text_list:
         char_list = []
-        text = text.translate(god_knows_why_en_testset_contains_zh_quote)
         text = text.translate(custom_trans)
         for seg in jieba.cut(text):
             seg_byte_len = len(bytes(seg, "UTF-8"))
@@ -150,22 +157,21 @@ def convert_char_to_pinyin(text_list, polyphone=True):
                 if char_list and seg_byte_len > 1 and char_list[-1] not in " :'\"":
                     char_list.append(" ")
                 char_list.extend(seg)
-            elif polyphone and seg_byte_len == 3 * len(seg):  # if pure chinese characters
-                seg = lazy_pinyin(seg, style=Style.TONE3, tone_sandhi=True)
-                for c in seg:
-                    if c not in "。，、；：？！《》【】—…":
+            elif polyphone and seg_byte_len == 3 * len(seg):  # if pure east asian characters
+                seg_ = lazy_pinyin(seg, style=Style.TONE3, tone_sandhi=True)
+                for i, c in enumerate(seg):
+                    if is_chinese(c):
                         char_list.append(" ")
-                    char_list.append(c)
-            else:  # if mixed chinese characters, alphabets and symbols
+                    char_list.append(seg_[i])
+            else:  # if mixed characters, alphabets and symbols
                 for c in seg:
                     if ord(c) < 256:
                         char_list.extend(c)
+                    elif is_chinese(c):
+                        char_list.append(" ")
+                        char_list.extend(lazy_pinyin(c, style=Style.TONE3, tone_sandhi=True))
                     else:
-                        if c not in "。，、；：？！《》【】—…":
-                            char_list.append(" ")
-                            char_list.extend(lazy_pinyin(c, style=Style.TONE3, tone_sandhi=True))
-                        else:  # if is zh punc
-                            char_list.append(c)
+                        char_list.append(c)
         final_text_list.append(char_list)
 
     return final_text_list