Update app.py

app.py

@@ -36,6 +36,9 @@ def augment_and_extract_features(audio_path, sr=16000, n_mfcc=40, n_fft=2048, ho
 def predict_voice(audio_file_path):
     try:
         features_tensor = augment_and_extract_features(audio_file_path)
+        # Reshape tensor to match model input size, avoiding dimension mismatch errors
+        if features_tensor.shape[-1] != model.config.num_labels:
+            features_tensor = torch.nn.functional.interpolate(features_tensor, size=(model.config.num_labels,), mode='linear', align_corners=False)
         with torch.no_grad():
             outputs = model(features_tensor)
 
@@ -60,3 +63,4 @@ iface = gr.Interface(
 )
 
 iface.launch()
+