Disable speech generation to stable the demo

app.py

@@ -57,7 +57,7 @@ from ola.utils import disable_torch_init
 from ola.datasets.preprocess import tokenizer_image_token, tokenizer_speech_image_token, tokenizer_speech_question_image_token
 from ola.mm_utils import get_model_name_from_path, KeywordsStoppingCriteria, process_anyres_video, process_anyres_highres_image_genli
 from ola.constants import IGNORE_INDEX, DEFAULT_IMAGE_TOKEN, IMAGE_TOKEN_INDEX, DEFAULT_SPEECH_TOKEN
-from ola.CosyVoice_main.cosyvoice.cli.cosyvoice import CosyVoice
+# from ola.CosyVoice_main.cosyvoice.cli.cosyvoice import CosyVoice
 
 from huggingface_hub import hf_hub_download
 
@@ -81,7 +81,7 @@ model = model.bfloat16()
 
 # tts_model = CosyVoice('iic/CosyVoice-300M-SFT', load_jit=False, fp16=True)
 # tts_model = CosyVoice('FunAudioLLM/CosyVoice-300M-SFT', load_jit=True, fp16=True)
-OUTPUT_SPEECH = False
+OUTPUT_SPEECH=False
 
 USE_SPEECH=False