MuLaCover enables flexible cover song creation with tonal control
MuLaCover: Flexible Cover Song Generation via Symbolic Tonal Control
Sound
Summary
Cover songs keep the melody and chords from a song but change other parts like style or tempo. The authors developed MuLaCover, which uses a music language model that follows a simplified music outline, called a lead sheet, instead of being locked to exact timing details. This lets users create cover songs that keep the same musical notes but vary the tempo, structure, and style freely. The authors trained MuLaCover to be more flexible and tested it against other tools, finding it performs very well.
What this means in practice
- •For music producers: Produce cover versions of songs by controlling melody and chords while freely changing timing and style.
- •For mobile app developers: Integrate flexible cover song generation features into music creation apps without requiring exact timing alignment.
Authors
Lynn Yi, Zihan Xiong, Junjie Cao, Xiaolong Weng, Yingzhe Ma, Gus Xia, Jinliang Liu, Yuxin Xie, Xianwei Zhuang, Yuguo Yin, Qiao Jin, Gongxi Zhu, Jiayu Wang, Junjie Liang, Qin Yue, Ziyu Wang, Dading Chong, Meng Cao, Jiahuan Zhou, Dongchao Yang
Abstract
A cover song re-renders an existing piece, most commonly by preserving its tonal content -- melody and chord progression -- while reshaping other musical attributes. This suggests a natural formulation for generative cover synthesis with pretrained music language models: explicitly control tonal content, while specifying lyrics and style through text. Existing approaches condition on frame-level features, tightly coupling generation to the reference's timing and structure, and limiting flexibility such as tempo variation, structural rearrangement, or partial conditioning. To this end, we present MuLaCover, a cover generation model that conditions a pretrained text-to-song foundation model on an editable symbolic lead sheet. Our design enables alignment-free control: generated outputs preserve the tonal signature of the source without frame-level alignment, allowing flexible timing, structure, and segment-level generation. We further introduce a training curriculum with partially mismatched audio-symbolic pairs, improving diversity and robustness. We evaluate MuLaCover with objective metrics, standard subjective ratings, and in-depth expert interviews that provide fine-grained diagnostic insights. Experiments show that MuLaCover achieves state-of-the-art performance among open-source systems and is competitive with leading commercial models such as Suno-v5.5. Audio samples and code are available at our project page: https://mulalabs.ai/mulacover/