Enhancing the vocal range of single-speaker singing voice synthesis with melody-unsupervised pre-training