arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2007.14974eess.AScs.SD
收藏

基于GAN的语音增强系统的损失函数和递归训练

On Loss Functions and Recurrency Training for GAN-based Speech Enhancement Systems

Didi AI Labs(滴滴人工智能实验室)

Zhuohuang Zhang, Chengyun Deng, Yi Shen, Donald S. Williamson, Yongtao Sha, Yi Zhang, Hui Song, Xiangang Li

摘要:

最近的工作表明,使用生成对抗网络(GANs)进行语音增强是可行的,然而,这些方法尚未与最先进的(SOTA)非基于GAN的方法进行比较。此外,已经为基于GAN的方法提出了许多损失函数,但它们还没有得到充分的比较。在这项研究中,我们提出了新的卷积递归GAN(CRGAN)架构的语音增强。采用多个损失函数,以便与其他基于GAN的系统进行直接比较。还探讨了包括经常性层的好处。我们的研究结果表明,提出的CR GAN模型优于SOTA GAN模型使用相同的损失函数,它优于其他非GAN为基础的系统,表明使用GAN的语音增强的好处。总体而言,结合客观度量损失函数与均方误差(MSE)的CRGAN模型在许多评估度量的比较方法上提供了最佳性能。

英文摘要:

Recent work has shown that it is feasible to use generative adversarial networks (GANs) for speech enhancement, however, these approaches have not been compared to state-of-the-art (SOTA) non GAN-based approaches. Additionally, many loss functions have been proposed for GAN-based approaches, but they have not been adequately compared. In this study, we propose novel convolutional recurrent GAN (CRGAN) architectures for speech enhancement. Multiple loss functions are adopted to enable direct comparisons to other GAN-based systems. The benefits of including recurrent layers are also explored. Our results show that the proposed CRGAN model outperforms the SOTA GAN-based models using the same loss functions and it outperforms other non-GAN based systems, indicating the benefits of using a GAN for speech enhancement. Overall, the CRGAN model that combines an objective metric loss function with the mean squared error (MSE) provides the best performance over comparison approaches across many evaluation metrics.

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。

↑