Update app.py

app.py

@@ -94,10 +94,10 @@ def custom_split_text(text: str) -> list:
             candidate_end = len(words)
         chunk_words = words[start:candidate_end]
         split_index = None
-        for i in range(len(chunk_words) - 1):
-            if '.' in chunk_words[i]:
-                split_index = i
-                break
+        # for i in range(len(chunk_words) - 1):
+        #     if '.' in chunk_words[i]:
+        #         split_index = i
+        #         break
         if split_index is not None:
             candidate_end = start + split_index + 1
             chunk_words = words[start:candidate_end]
@@ -198,7 +198,7 @@ def tts_streaming(text: str, voice: str = "af_heart", speed: float = 1.0, format
             if i == 0:
                 tokens_to_send = [0] + chunk_tokens + [0]
             else:
-                tokens_to_send = [0] + [prev_last_token] + [16] + chunk_tokens + [0]
+                tokens_to_send = [prev_last_token] + [16] + chunk_tokens + [0]
             #     token_to_send = [0] + chunk_tokens + [0]
 
             # Save the last token of this chunk for the next iteration.
@@ -209,7 +209,7 @@ def tts_streaming(text: str, voice: str = "af_heart", speed: float = 1.0, format
             print(final_token)
 
             # Use the number of tokens to select the appropriate style vector.
-            style_index = len(chunk_tokens) + 2
+            style_index = len(chunk_tokens) + 1
             if style_index >= len(voices):
                 style_index = len(voices) - 1  # Fallback if index is out-of-bounds.
             ref_s = voices[style_index]