Speech enhancement with variational autoencoders and alpha-stable distributions