How can I compute the AUC by using Gaussian Mixture Model?

By using this code, can I compute the AUC:

clf=GaussianMixture(n_components=3).fit(X_train)
scores=clf.score_samples(X_test)
scores=np.exp(scores)
fpr, tpr, _=roc_curve(Y_trut, scores)# scores are the PDF 
roc_auc = auc(fpr, tpr)

For the roc_curve function, is it correct to pass scores as probability density or probability?