gradio infer

MMAudio/demo.py

@@ -64,7 +64,7 @@ def v2a_load():
     if args.variant not in all_model_cfg:
         raise ValueError(f'Unknown model variant: {args.variant}')
     model: ModelConfig = all_model_cfg[args.variant]
-    #model.download_if_needed()
+    model.download_if_needed()
     seq_cfg = model.seq_cfg
 
     #if args.video:

app.py

@@ -35,11 +35,6 @@ if True:
 log = logging.getLogger()
 
 
-import sys
-sys.path.insert(0, "./F5-TTS/src/")
-from f5_tts.infer.infer_cli_test import v2s_infer
-
-
 import sys
 sys.path.insert(0, "./MMAudio/")
 from demo import v2a_load, v2a_infer
@@ -47,6 +42,11 @@ from demo import v2a_load, v2a_infer
 v2a_loaded = v2a_load()
 
 
+import sys
+sys.path.insert(0, "./F5-TTS/src/")
+from f5_tts.infer.infer_cli_test import v2s_infer
+
+
 #@spaces.GPU(duration=120)
 def video_to_audio_and_speech(video: gr.Video, prompt: str, v2a_num_steps: int, text: str, audio_prompt: gr.Audio, text_prompt: str, v2s_num_steps: int):