LeVo: High-Quality Song Generation with Multi-Preference Alignment

Recent advances in large language models (LLMs) and audio language models have significantly improved music generation, particularly in lyrics-to-song generation.