trying things with cpu

app.py

@@ -34,7 +34,7 @@ from huggingface_hub import hf_hub_download
 from llama_cpp import Llama
 from TTS.api import TTS
 from TTS.utils.manage import ModelManager
-#from TTS.tts.configs.xtts_config import XttsConfig
+from TTS.tts.configs.xtts_config import XttsConfig
 #from TTS.tts.models.xtts import Xtts
 #from TTS.utils.generic_utils import get_user_data_dir
 #from TTS.utils.manage import ModelManager
@@ -154,7 +154,7 @@ with gr.Blocks(title="Voice chat with LLM") as demo:
         def handle_speech_generation(sentence, chatbot_history, chatbot_voice):
             if sentence != "":
                 print("Processing sentence")
-                generated_speech = generate_speech_for_sentence(chatbot_history, chatbot_voice, sentence, xtts_model, xtts_supported_languages=config.languages, return_as_byte=True)
+                generated_speech = generate_speech_for_sentence(chatbot_history, chatbot_voice, sentence, xtts_model, None, return_as_byte=True)
                 if generated_speech is not None:
                     _, audio_dict = generated_speech
                     yield (sentence, chatbot_history, audio_dict["value"])