Update app.py

app.py

@@ -93,20 +93,35 @@ def _inference_forward_stream(
         yield spectrogram
 
 
-def get_model(name_model):
+def  get_model(name_model):
     global models
-    if name_model in models:
-        tokenizer = AutoTokenizer.from_pretrained(name_model, token=token)
-        return models[name_model], tokenizer
+    if name_model in   models:
+        if  name_model=='wasmdashai/vits-en-v1':
+            tokenizer = AutoTokenizer.from_pretrained("wasmdashai/vits-en-v1",token=token)
+        else: 
+            tokenizer = AutoTokenizer.from_pretrained("wasmdashai/vtk",token=token)
 
-    models[name_model] = VitsModel.from_pretrained(name_model, token=token)
+        
+
+        
+        return models[name_model],tokenizer
+    models[name_model]=VitsModel.from_pretrained(name_model,token=token)
+    
+    
+   
     models[name_model].decoder.apply_weight_norm()
+    # torch.nn.utils.weight_norm(self.decoder.conv_pre)
+    # torch.nn.utils.weight_norm(self.decoder.conv_post)
     for flow in models[name_model].flow.flows:
         torch.nn.utils.weight_norm(flow.conv_pre)
         torch.nn.utils.weight_norm(flow.conv_post)
+    
+    if  name_model=='wasmdashai/vits-en-v1':
+            tokenizer = AutoTokenizer.from_pretrained("wasmdashai/vits-en-v1",token=token)
+    else: 
+            tokenizer = AutoTokenizer.from_pretrained("wasmdashai/vtk",token=token)
 
-    tokenizer = AutoTokenizer.from_pretrained(name_model, token=token)
-    return models[name_model], tokenizer
+    return models[name_model],tokenizer
 
 TXT = """السلام عليكم ورحمة الله وبركاته يا هلا وسهلا ومراحب بالغالي اخباركم طيبين ان شاء الله ارحبوا على العين والراس"""
 def process_chunk(chunk_id, spectrogram_chunk, speaker_embeddings, decoder):