engine = pyttsx3.init()

app.py

@@ -116,22 +116,33 @@ def main_function(password: str, theme: str, language: str, method: str, rounds:
 
     return chatbot_dialogue, audio_path, file_name
 
-
 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']} "
+    engine = pyttsx3.init()
+
+    # Fetch the list of available voices
+    voices = engine.getProperty('voices')
+
+    # Get voice IDs for male and female voices (you might need to adjust these based on available voices on your system)
+    male_voice_id = "com.apple.speech.synthesis.voice.alex"  # Example ID for a male voice
+    female_voice_id = "com.apple.speech.synthesis.voice.victoria"  # Example ID for a female voice
     
-    tts = gTTS(text=text, lang='zh-tw')
     file_path = "temp_audio.mp3"
-    tts.save(file_path)
+    
+    for i, item in enumerate(dialogue):
+        gender = role1_gender if i % 2 == 0 else role2_gender
+        voice_id = male_voice_id if gender == "male" else female_voice_id
+        
+        # Set the voice
+        engine.setProperty('voice', voice_id)
+
+        # Now, synthesize the speech
+        engine.save_to_file(item['content'], file_path)
+        engine.runAndWait()
+
     return file_path
 
 
+
 if __name__ == "__main__":
     gr.Interface(
         main_function, 

requirements.txt

@@ -1,3 +1,4 @@
 gradio
 gtts
-openai
+openai
+pyttsx3