A Study of Learning Based Beamforming Methods for Speech Recognition

By Xiong Xiao, Chenglin Xu, Zhaofeng Zhang, Shengkui Zhao, Sining Sun, Shinji Watanabe, Longbiao Wang, Lei Xie, Douglas L. Jones, E. Chng and 1 more. 2016

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Eliya Nachmani
2. Yossi Adi
3. Lior Wolf
4. Nima Mesgarani
5. Yi Luo
6. Cong Han
7. Alexandre D'efossez
8. Wei-Ning Hsu

Citations in the corpus

• Voice Separation with an Unknown Number of Multiple Speakers (2020, methodology): “The MVDR neural beamformer introduce in (Xiao et al., 2016), predict a mask in time-frequency space, which is then used to estimate the MVDR beam, showing an improvement over both the traditional MVDR method and delay-and-sum beamforming.”
• FaSNet: Low-Latency Adaptive Beamforming for Multi-Microphone Audio Processing (2019, methodology): “With the dominance of deep learning methods in almost all audio processing tasks, deep learning-based beamformers, sometimes called neural beamformers , have also proven effective especially when jointly trained together with backend models for tasks such as automatic speech recognition (ASR) [2].”