Monaural speech enhancement has made dramatic advances since the introduction of deep learning a few years ago. Although enhanced speech has been demonstrated to have better intelligibility and quality for human listeners, feeding it directly to automatic speech recognition (ASR) systems trained with noisy speech has not produced expected improvements in ASR performance. The lack of an enhancement benefit on recognition, or the gap between monaural speech enhancement and recognition, is often attributed to speech distortions introduced in the enhancement process. In this study, we analyze the distortion problem, compare different acoustic models, and investigate a distortion-independent training scheme for monaural speech recognition. Experimental results suggest that distortion-independent acoustic modeling is able to overcome the distortion problem. Such an acoustic model can also work with speech enhancement models different from the one used during training. Moreover, the models investigated in this paper outperform the previous best system on the CHiME-2 corpus.
Utterance Weighted Multi-Dilation Temporal Convolutional Networks for Monaural Speech Dereverberation
The University of Sheffield（谢菲尔德大学）
语音去混响是许多语音技术应用中的一个重要阶段。最近在这一领域的工作一直由深度神经网络模型主导。时间卷积网络（TCN）是深度学习模型，已被提出用于语音去混响任务中的序列建模。在这项工作中，加权多膨胀深度可分离卷积提出了取代标准的深度可分离卷积TCN模型。这种建议的卷积使TCN能够动态地关注网络中每个卷积块的感受野中或多或少的局部信息。结果表明，这种加权多膨胀时间卷积网络（WD-TCN）在各种模型配置中始终优于TCN，并且使用WD-TCN模型是一种比增加卷积块数量更有效的参数方法来提高模型的性能。与基线TCN相比，最佳性能改善为0.55 dB尺度不变信号失真比（SISDR），而WD-TCN模型在WHAMR数据集上的最佳性能达到12.26 dB SISDR。
Reducing the Gap Between Pretrained Speech Enhancement and Recognition Models Using a Real Speech-Trained Bridging Module
Tianjin University（天津大学）；DiDi Chuxing（滴滴出行）；Kyoto University（京都大学）；Huiyan Technology (Tianjin) Co., Ltd（慧研科技（天津）有限公司）；Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences（中国科学院深圳先进技术研究院）
针对预训练语音增强与 ASR 模型间因模拟训练数据不匹配导致的性能损失，提出用 DNSMOS 和多任务 WER 向量在真实带噪语音上训练 OA 桥接模块，在 CHiME-4 上显著降低 WER。
School of Computer Science, Wuhan University（武汉大学计算机学院）；Suzhou Municipal Key Laboratory of Multimodal Intelligent Systems, Digital Innovation Research Center, Duke Kunshan University（多模态智能系统苏州市重点实验室、创新数字研究中心、杜克昆山大学）
