Update app.py

app.py

@@ -192,19 +192,20 @@ def tts_streaming(text: str, voice: str = "af_heart", speed: float = 1.0, format
         for i, chunk in enumerate(chunks):
             # Convert the chunk text to tokens.
             chunk_tokens = tokenizer(chunk)
+            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
-                token_to_send = [0] + chunk_tokens + [0]
+            # if i == 0:
+            #     tokens_to_send = [0] + chunk_tokens + [0]
+            # else:
+            #     # tokens_to_send = [prev_last_token] + [0] + chunk_tokens
+            #     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 = tokens_to_send[-1]
 
             # Prepare the model input (a batch of one sequence).
-            final_token = [tokens_to_send]
+            final_token = [[0, *chunk_tokens, 0]]
 
             # Use the number of tokens to select the appropriate style vector.
             style_index = len(tokens_to_send)