arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2312.01744eess.AS
收藏

SEFGAN:利用归一化流与GAN的力量实现高效高质量语音增强

SEFGAN: Harvesting the Power of Normalizing Flows and GANs for Efficient High-Quality Speech Enhancement

Martin Strauss, Nicola Pia, Nagashree K. S. Rao, Bernd Edler

AI总结:

本文提出SEFGAN,将归一化流作为GAN生成器并结合最大似然训练用于语音增强,在不增加复杂度的情况下显著优于NF基线,并与最先进模型相竞争。

摘要:

本文提出了SEFGAN,这是一种结合最大似然训练和生成对抗网络(GANs)的深度神经网络(DNN),用于高效语音增强(SE)。为此,训练一个DNN,以含噪语音为条件,在GAN框架中使用归一化流(NF)作为生成器来合成增强语音。虽然似然模型与GAN的结合并非易事,但SEFGAN证明了混合对抗训练和最大似然训练的方法能够使模型同时保持高质量的音频生成和对数似然估计。我们的实验表明,该方法在不增加增强网络额外复杂度的情况下,显著优于基于NF的基线模型。使用计算指标和听力实验进行的比较表明,SEFGAN与其他最先进的模型相比具有竞争力。

英文摘要:

This paper proposes SEFGAN, a Deep Neural Network (DNN) combining maximum likelihood training and Generative Adversarial Networks (GANs) for efficient speech enhancement (SE). For this, a DNN is trained to synthesize the enhanced speech conditioned on noisy speech using a Normalizing Flow (NF) as generator in a GAN framework. While the combination of likelihood models and GANs is not trivial, SEFGAN demonstrates that a hybrid adversarial and maximum likelihood training approach enables the model to maintain high quality audio generation and log-likelihood estimation. Our experiments indicate that this approach strongly outperforms the baseline NF-based model without introducing additional complexity to the enhancement network. A comparison using computational metrics and a listening experiment reveals that SEFGAN is competitive with other state-of-the-art models.

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。

↑