Update src/utils.py

src/utils.py

@@ -7,7 +7,7 @@ from typing import Iterator, TextIO, Union
 import tqdm
 
 import urllib3
-
+import opencc
 
 def exact_div(x, y):
     assert x % y == 0
@@ -52,8 +52,10 @@ def format_timestamp(seconds: float, always_include_hours: bool = False, fractio
 
 
 def write_txt(transcript: Iterator[dict], file: TextIO):
+    converter = opencc.OpenCC('s2twp') #s2twp.json - 簡體到繁體（臺灣正體標準）並轉換爲臺灣常用詞彙
     for segment in transcript:
-        print(segment['text'].strip(), file=file, flush=True)
+        segment_text = converter.convert(segment['text'].strip())
+        print(segment_text, file=file, flush=True)
 
 
 def write_vtt(transcript: Iterator[dict], file: TextIO, 
@@ -86,9 +88,11 @@ def write_srt(transcript: Iterator[dict], file: TextIO,
             write_srt(result["segments"], file=srt)
     """
     iterator  = __subtitle_preprocessor_iterator(transcript, maxLineWidth, highlight_words)
+    converter = opencc.OpenCC('s2twp') #s2twp.json - 簡體到繁體（臺灣正體標準）並轉換爲臺灣常用詞彙
 
     for i, segment in enumerate(iterator, start=1):
-        text = segment['text'].replace('-->', '->')
+        segment_text = segment['text'].replace('-->', '->')
+        text = converter.convert(segment_text)
 
         # write srt lines
         print(