def dialogue_to_audio(dialogue, role1_gender, role2_gender):

app.py

@@ -117,13 +117,21 @@ def main_function(password: str, theme: str, language: str, method: str, rounds:
     return chatbot_dialogue, audio_path, file_name
 
 
-def dialogue_to_audio(dialogue):
-    text = " ".join([f"{item['content']}" for item in dialogue])
+def dialogue_to_audio(dialogue, role1_gender, role2_gender):
+    # Assuming the dialogue follows the format of alternating roles.
+    text = ""
+    for i, item in enumerate(dialogue):
+        gender = role1_gender if i % 2 == 0 else role2_gender
+        voice_attribute = "male" if gender == "male" else "female"
+        # For simplicity, we'll just prepend the voice attribute. In a real-world scenario, you'd use this to select different TTS voices.
+        text += f"{voice_attribute} voice: {item['content']} "
+    
     tts = gTTS(text=text, lang='zh-tw')
     file_path = "temp_audio.mp3"
     tts.save(file_path)
     return file_path
 
+
 if __name__ == "__main__":
     gr.Interface(
         main_function,