The SJTU X-LANCE Lab System for MSR Challenge 2025

TL;DR

This paper presents a system for music source restoration that combines sequential BS-RoFormers for multiple tasks, achieving top rankings in the MSR Challenge 2025 with open-sourced code.

Contribution

The novel system integrates sequential BS-RoFormers for multi-task music source restoration and employs advanced training schemes, setting new performance benchmarks.

Findings

Achieved first place in all evaluation metrics.

Attained MMSNR score of 4.4623.

Achieved FAD score of 0.1988.

Abstract

This report describes the system submitted to the music source restoration (MSR) Challenge 2025. Our approach is composed of sequential BS-RoFormers, each dealing with a single task including music source separation (MSS), denoise and dereverb. To support 8 instruments given in the task, we utilize pretrained checkpoints from MSS community and finetune the MSS model with several training schemes, including (1) mixing and cleaning of datasets; (2) random mixture of music pieces for data augmentation; (3) scale-up of audio length. Our system achieved the first rank in all three subjective and three objective evaluation metrics, including an MMSNR score of 4.4623 and an FAD score of 0.1988. We have open-sourced all the code and checkpoints at https://github.com/ModistAndrew/xlance-msr.