Update web_demo.py

web_demo.py

@@ -25,6 +25,10 @@ audio_token_pattern = re.compile(r"<\|audio_(\d+)\|>")
 from flow_inference import AudioDecoder
 from audio_process import AudioStreamProcessor
 
+use_local_interface = True  
+if use_local_interface :
+    from model_server import ModelWorker 
+
 if __name__ == "__main__":
     parser = ArgumentParser()
     parser.add_argument("--host", type=str, default="0.0.0.0")
@@ -60,6 +64,13 @@ if __name__ == "__main__":
         whisper_model = WhisperVQEncoder.from_pretrained(args.tokenizer_path).eval().to(device)
         feature_extractor = WhisperFeatureExtractor.from_pretrained(args.tokenizer_path)
 
+        global use_local_interface, worker  
+        if  use_local_interface :
+            model_path0 = 'THUDM/glm-4-voice-9b'
+            # dtype = 'bfloat16'
+            device0 = 'cpu'
+            worker = ModelWorker(model_path0,device0) 
+
 
     def clear_fn():
         return [], [], '', '', '', None, None