中文标题#
语音分离中带有噪声参考的尺度不变信干比研究
英文标题#
A Study of the Scale Invariant Signal to Distortion Ratio in Speech Separation with Noisy References
中文摘要#
本文研究了在监督语音分离中，当训练参考包含噪声时（如事实上的基准数据集 WSJ0-2Mix），将尺度不变信噪比（SI-SDR）用作评估和训练目标的含义。对带有噪声参考的 SI-SDR 的推导表明，噪声会限制可达到的 SI-SDR，或导致分离输出中出现不需要的噪声。为了解决这个问题，提出了一种增强参考的方法，并使用 WHAM! 对混合信号进行增强，旨在训练避免学习噪声参考的模型。在这些增强数据集上训练的两个模型使用非侵入式 NISQA.v2 度量进行评估。结果表明分离语音中的噪声减少，但表明处理参考可能会引入伪影，限制整体质量的提升。在 WSJ0-2Mix 和 Libri2Mix 测试集上，所有模型的 SI-SDR 与感知噪声之间存在负相关，这支持了从推导中得出的结论。
英文摘要#
This paper examines the implications of using the Scale-Invariant Signal-to-Distortion Ratio (SI-SDR) as both evaluation and training objective in supervised speech separation, when the training references contain noise, as is the case with the de facto benchmark WSJ0-2Mix. A derivation of the SI-SDR with noisy references reveals that noise limits the achievable SI-SDR, or leads to undesired noise in the separated outputs. To address this, a method is proposed to enhance references and augment the mixtures with WHAM!, aiming to train models that avoid learning noisy references. Two models trained on these enhanced datasets are evaluated with the non-intrusive NISQA.v2 metric. Results show reduced noise in separated speech but suggest that processing references may introduce artefacts, limiting overall quality gains. Negative correlation is found between SI-SDR and perceived noisiness across models on the WSJ0-2Mix and Libri2Mix test sets, underlining the conclusion from the derivation.
文章页面#
PDF 获取#
抖音扫码查看更多精彩内容