Update app.py

app.py

@@ -195,21 +195,20 @@ def tts_streaming(text: str, voice: str = "af_heart", speed: float = 1.0, format
             print(chunk)
 
             # For the first chunk, prepend 0; for later chunks, start with the previous chunk's last token.
-            # if i == 0:
-            #     tokens_to_send = [0] + chunk_tokens + [0]
-            # else:
-            #     # tokens_to_send = [prev_last_token] + [0] + chunk_tokens
+            if i == 0:
+                tokens_to_send = [0] + chunk_tokens + [0]
+            else:
+                tokens_to_send = [prev_last_token] + chunk_tokens + [0]
             #     token_to_send = [0] + chunk_tokens + [0]
 
             # Save the last token of this chunk for the next iteration.
-            # prev_last_token = tokens_to_send[-1]
+            prev_last_token = chunk_tokens[-1]
 
             # Prepare the model input (a batch of one sequence).
-            final_token = [[0, *chunk_tokens, 0]]
+            final_token = [tokens_to_send]
 
             # Use the number of tokens to select the appropriate style vector.
-            style_index = len(final_token)
-            print(len(voices))
+            style_index = len(chunk_tokens) + 2
             if style_index >= len(voices):
                 style_index = len(voices) - 1  # Fallback if index is out-of-bounds.
             ref_s = voices[style_index]