The INTERSPEECH 2020 Deep Noise Suppression Challenge: Datasets, Subjective Testing Framework, and Challenge Results
Microsoft Corporation(微软公司)
Chandan K. A. Reddy, Vishak Gopal, Ross Cutler, Ebrahim Beyrami, Roger Cheng, Harishchandra Dubey, Sergiy Matusevych, Robert Aichner, Ashkan Aazami, Sebastian B… 展开作者
Chandan K. A. Reddy, Vishak Gopal, Ross Cutler, Ebrahim Beyrami, Roger Cheng, Harishchandra Dubey, Sergiy Matusevych, Robert Aichner, Ashkan Aazami, Sebastian Braun, Puneet Rana, Sriram Srinivasan, Johannes Gehrke
备注:
Interspeech 2020. arXiv admin note: substantial text overlap with arXiv:2001.08662
The INTERSPEECH 2020 Deep Noise Suppression (DNS) Challenge is intended to promote collaborative research in real-time single-channel Speech Enhancement aimed to maximize the subjective (perceptual) quality of the enhanced speech. A typical approach to evaluate the noise suppression methods is to use objective metrics on the test set obtained by splitting the original dataset. While the performance is good on the synthetic test set, often the model performance degrades significantly on real recordings. Also, most of the conventional objective metrics do not correlate well with subjective tests and lab subjective tests are not scalable for a large test set. In this challenge, we open-sourced a large clean speech and noise corpus for training the noise suppression models and a representative test set to real-world scenarios consisting of both synthetic and real recordings. We also open-sourced an online subjective test framework based on ITU-T P.808 for researchers to reliably test their developments. We evaluated the results using P.808 on a blind test set. The results and the key learnings from the challenge are discussed. The datasets and scripts can be found here for quick access https://github.com/microsoft/DNS-Challenge.
DPT-FSNet: Dual-path Transformer Based Full-band and Sub-band Fusion Network for Speech Enhancement
Institute of Acoustics, CAS(中国科学院声学研究所);University of Chinese Academy of Sciences(中国科学院大学)
子带模型已经取得了可喜的成果,由于他们的能力,在频谱图中的局部模式建模。一些研究通过融合子带和全带信息进一步提高性能。然而,全频带和子频带融合模型的结构没有被充分探索。提出了一种基于双通道变换器的全带与子带融合网络(DPT-FSNet)用于频域语音增强。双路径Transformer的帧内和帧间部分分别模拟子带和全带信息。我们所提出的方法所利用的功能是更可解释的时域双路Transformer所利用的。我们在Voice Bank + DEMAND和Interspeech 2020深度噪声抑制(DNS)数据集上进行了实验,以评估所提出的方法。实验结果表明,该方法优于目前的国家的最先进的。
复频谱和幅度被认为是语音增强和去混响的两个主要特征。传统的方法总是将这两个特征分开处理,忽略了它们之间的内在联系。本文提出了Uformer,一种基于Unet的复-实双径扩张保形网络,用于同时进行语音增强和混响消除。我们利用时间注意力(TA)和扩张卷积(DC)来利用局部和全局上下文信息和频率注意力(FA)来建模维度信息。这三个子模块组成的扩展复实双径一致器模块有效地提高了语音增强和去混响性能。此外,采用混合编码器和解码器,同时对复频谱和幅度进行建模,促进两个域之间的信息交互。编码器/解码器的注意力也被应用于增强编码器和解码器之间的交互。我们的实验结果优于所有的SOTA时间和复杂的域模型客观和主观。具体而言,Uformer在Interspeech 2021 DNS Challenge的盲测集上达到3.6032 DNSMOS,优于所有表现最好的模型。我们还进行了烧蚀实验,梳理除了所有提出的子模块是最重要的。
A Simultaneous Denoising and Dereverberation Framework with Target Decoupling
Key Laboratory of Noise and Vibration Research, Institute of Acoustics, Chinese Academy of Sciences(中国科学院声学研究所噪声与振动重点实验室);University of Chinese Academy of Sciences(中国科学院大学);Communication University of China(中国传媒大学)
Does a PESQNet (Loss) Require a Clean Reference Input? The Original PESQ Does, But ACR Listening Tests Don't
语音质量的感知评价(PESQ)需要一个干净的语音参考作为输入,但预测的结果(无参考)绝对类别评级(ACR)测试。在这项工作中,我们训练一个完全卷积的递归神经网络(FCRN)作为深度噪声抑制(DNS)模型,使用非侵入式或侵入式PESQNet,其中只有后者可以访问干净的语音参考。PESQNet被用作在DNS训练期间提供感知损失的中介,以最大化增强语音信号的PESQ分数。对于侵入式PESQNet,我们研究了两种拓扑结构,称为早期融合(EF)和中期融合(MF)PESQNet,并与非侵入式PESQNet进行比较,以评估和量化在DNS训练期间采用干净语音参考输入的好处。详细的分析表明,使用MF侵入式PESQNet训练的DNS分别比Interspeech 2021 DNS Challenge基线和使用MSE损失训练的相同DNS分别高出0.23和0.12个PESQ点。此外,我们可以证明,与使用非侵入式PESQNet训练的DNS相比,只获得了边际效益。因此,作为ACR听力测试,PESQNet不一定需要干净的语音参考输入,从而可以使用真实数据进行DNS训练。
CleanUMamba: A Compact Mamba Network for Speech Denoising using Channel Pruning
Faculty of Electrical Engineering, Mathematics and Computer Science, Delft University of Technology(电气工程、数学和计算机科学学院,代尔夫特理工大学);Leiden Institute of Advanced Computer Science (LIACS), Leiden University(先进计算机科学研究所(LIACS),莱顿大学)
Developing a Top-tier Framework in Naturalistic Conditions Challenge for Categorized Emotion Prediction: From Speech Foundation Models and Learning Objective to Data Augmentation and Engineering Choices
Non-parallel Voice Conversion System with WaveNet Vocoder and Collapsed Speech Suppression
Graduate School of Informatics, Nagoya University(名古屋大学信息学研究科);Information Technology Center, Nagoya University(名古屋大学信息技术中心);Graduate School of Information Science, Nagoya University(名古屋大学信息科学研究科)