arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2310.01809cs.SDeess.AS
收藏

用于音乐源分离的Mel-Band RoFormer

Mel-Band RoFormer for Music Source Separation

ByteDance(字节跳动)

Ju-Chiang Wang, Wei-Tsung Lu, Minz Won

AI总结:

针对现有BS-RoFormer音乐源分离模型的经验式无重叠波段划分缺乏理论支撑的问题,提出基于梅尔标度的重叠式梅尔波段划分的Mel-RoFormer模型,在MUSDB18HQ数据集上多音轨分离性能优于原有模型。

摘要:

近期,基于多波段频谱图的方法(如Band-Split RNN,即BSRNN)已在音乐源分离任务中展现出可观效果。在我们此前的工作中,提出了BS-RoFormer模型,该模型在前端继承了BSRNN的波段划分思路,随后采用带旋转位置嵌入(RoPE)的分层Transformer对带内和带间序列建模,以实现多波段掩膜估计。该模型虽取得了当前最优性能,但其波段划分方案是凭经验定义的,缺乏文献中的分析支撑。本文提出Mel-RoFormer,采用梅尔波段方案,即根据梅尔标度将频率仓映射到有重叠的子波段中;相比之下,BSRNN和BS-RoFormer中的波段划分映射是无重叠且基于启发式规则设计的。我们使用MUSDB18HQ数据集开展实验,结果表明Mel-RoFormer在人声、鼓及其他音轨的分离任务上性能优于BS-RoFormer。

英文摘要:

Recently, multi-band spectrogram-based approaches such as Band-Split RNN (BSRNN) have demonstrated promising results for music source separation. In our recent work, we introduce the BS-RoFormer model which inherits the idea of band-split scheme in BSRNN at the front-end, and then uses the hierarchical Transformer with Rotary Position Embedding (RoPE) to model the inner-band and inter-band sequences for multi-band mask estimation. This model has achieved state-of-the-art performance, but the band-split scheme is defined empirically, without analytic supports from the literature. In this paper, we propose Mel-RoFormer, which adopts the Mel-band scheme that maps the frequency bins into overlapped subbands according to the mel scale. In contract, the band-split mapping in BSRNN and BS-RoFormer is non-overlapping and designed based on heuristics. Using the MUSDB18HQ dataset for experiments, we demonstrate that Mel-RoFormer outperforms BS-RoFormer in the separation tasks of vocals, drums, and other stems.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑