论文标题
多任务深度残留回声抑制和回声损失
Multi-Task Deep Residual Echo Suppression with Echo-aware Loss
论文作者
论文摘要
本文介绍了NWPU团队参加ICASSP 2022 AEC挑战的参赛作品。我们采用了一种混合方法，该方法将带有神经后过滤器的线性AEC级联。前者用于处理线性回声组件，而后者抑制了残留的非线性回声组件。我们使用封闭式卷积F-T-LSTM神经网络（GFTNN）作为骨干，并通过多任务学习（MTL）框架来塑造后过滤器，在该框架中，语音活动检测（VAD）模块被用作辅助任务，并避免过度抑制言语膨胀。此外，我们采用回声感知损失函数，可以根据信噪比（SER）对均方根误差（MSE）损失进行优化，尤其是针对每个时频箱（TF-bin），从而进一步抑制了ECHO。广泛的消融研究表明，神经后过滤器中的时间延迟估计（TDE）模块会带来更好的感知质量，并且具有更好收敛性的自适应过滤器将为后过滤器带来一致的性能增长。此外，我们发现使用线性回波作为神经后过滤器的输入比直接使用参考信号更好。在ICASSP 2022 AEC-CHALLENGE中，我们的方法在单词准确度（WACC）（0.817）和平均意见分数（MOS）（4.502）和最终得分（0.864）上排名第一（0.817）（0.817）。
This paper introduces the NWPU Team's entry to the ICASSP 2022 AEC Challenge. We take a hybrid approach that cascades a linear AEC with a neural post-filter. The former is used to deal with the linear echo components while the latter suppresses the residual non-linear echo components. We use gated convolutional F-T-LSTM neural network (GFTNN) as the backbone and shape the post-filter by a multi-task learning (MTL) framework, where a voice activity detection (VAD) module is adopted as an auxiliary task along with echo suppression, with the aim to avoid over suppression that may cause speech distortion. Moreover, we adopt an echo-aware loss function, where the mean square error (MSE) loss can be optimized particularly for every time-frequency bin (TF-bin) according to the signal-to-echo ratio (SER), leading to further suppression on the echo. Extensive ablation study shows that the time delay estimation (TDE) module in neural post-filter leads to better perceptual quality, and an adaptive filter with better convergence will bring consistent performance gain for the post-filter. Besides, we find that using the linear echo as the input of our neural post-filter is a better choice than using the reference signal directly. In the ICASSP 2022 AEC-Challenge, our approach has ranked the 1st place on word accuracy (WAcc) (0.817) and the 3rd place on both mean opinion score (MOS) (4.502) and the final score (0.864).
