Update app.py

app.py

@@ -632,18 +632,21 @@ def generate_story_and_speech(input_text, chatbot_role):
     if last_history is not None:
         # Convert the list of lists back into a list of tuples for the history
         history_tuples = [tuple(entry) for entry in last_history]
-        synthesized_speech = generate_speech_for_sentence(history_tuples, chatbot_role, story_text) 
+
+        synthesized_speech = generate_speech_for_sentence(history_tuples, chatbot_role, story_text, return_as_byte=True) 
         if synthesized_speech:
+            # Get the Gradio Audio object
+            audio_obj = synthesized_speech[1]
             # Access the BytesIO object containing the WAV file and extract bytes
-            speech_audio_bytes = synthesized_speech[1]["value"]  # Assume speech audio is always in bytearray format
-            
+            speech_audio_bytes = audio_obj.data  # Use the 'data' attribute to get the bytearray
             # Convert the speech audio bytes to base64 for JSON serialization
             speech_audio_base64 = base64.b64encode(speech_audio_bytes).decode('utf8')
-
-
             return {"text": story_text.strip(), "audio": speech_audio_base64}
+        else:
+            return {"text": "Failed to generate story (no synthesized speech)", "audio": None}
+    
     else:
-        return {"text": "Failed to generate story", "audio": None}
+        return {"text": "Failed to generate story (last_history is empty)", "audio": None}
 
 # Create a Gradio Interface using only the `generate_story_and_speech()` function and the 'json' output type
 demo = gr.Interface(