An Empirical Study of Speech Language Models for Prompt-Conditioned Speech Synthesis