diff --git a/src/CLAPWrapper.py b/src/CLAPWrapper.py
index d69d3bf..7db3622 100644
--- a/src/CLAPWrapper.py
+++ b/src/CLAPWrapper.py
@@ -149,7 +149,7 @@ class CLAPWrapper():
             audio_tensor = self.load_audio_into_tensor(
                 audio_file, self.args.duration, resample)
             audio_tensor = audio_tensor.reshape(
-                1, -1).cuda if self.use_cuda and torch.cuda.is_available() else audio_tensor.reshape(1, -1)
+                1, -1).cuda() if self.use_cuda and torch.cuda.is_available() else audio_tensor.reshape(1, -1)
             audio_tensors.append(audio_tensor)
         return self.default_collate(audio_tensors)