Improve the multilingual parser

app.py

@@ -32,9 +32,14 @@ text_examples = [
 ]
 
 def parse_multilingual_text(input_text):
-    pattern = r"<(\w+)>\s(.*?)\s(?=<\w+>|$)"
-    segments = re.findall(pattern, input_text)
-    return [(lang, text.strip()) for lang, text in segments if lang in LANGUAGES.keys()]
+    pattern = r"(?:<(\w+)>)|([^<]+)"
+    cur_lang = 'en'
+    segments = []
+    for i, (lang, txt) in enumerate(re.findall(pattern, input_text)):
+        if lang: cur_lang = lang
+        else: segments.append((cur_lang, f"  {txt}  ")) # add spaces to give it some time to switch languages
+    if not segments: return [("en", "")]
+    return segments
 
 @spaces.GPU(enable_queue=True)
 def generate_segment_audio(text, lang, speaker_audio, pipe):