Speech Dereverberation Based on Variance-Normalized Delayed Linear Prediction
By Tomohiro Nakatani, Takuya Yoshioka, Keisuke Kinoshita, M. Miyoshi, Biing-Hwang Fred Juang. 2010
477 citations recorded by the source corpus.
Subfields: Computational Mechanics, Electrical and Electronic Engineering, Speech & Audio, Signal Processing
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computational Mechanics
- Shinji Watanabe
- Sharon Gannot
- Jens Heitkaemper
- Reinhold Haeb-Umbach
- Naoyuki Kanda
- Peter Gerstoft
- Christoph Boeddeker
- Charles-Alban Deledalle
In Electrical and Electronic Engineering
In Speech & Audio
- Shinji Watanabe
- Yusuke Fujita
- Shota Horiguchi
- Sanjeev Khudanpur
- Naoyuki Kanda
- Kenji Nagamatsu
- Vimal Manohar
- Christoph Boeddeker
In Signal Processing
- Shinji Watanabe
- Yusuke Fujita
- Shota Horiguchi
- Jinyu Li
- Naoyuki Kanda
- Kenji Nagamatsu
- Vimal Manohar
- Sanjeev Khudanpur
Citations in the corpus
- Front-end processing for the CHiME-5 dinner party scenario (2018, methodology): “Additionally, we apply multi-channel dereverberation based on the Weighted Prediction Error (WPE) principle [14, 15, 16] as Both authors contributed equally. a preprocessing step for GSS.”
- Machine learning in acoustics: Theory and applications. (2019, background): “This can be exploited to yield a dereverberation algo-rithm, 250,258–264 and to estimate signal direction-of-arrival.”
- NARA-WPE: A Python package for weighted prediction error dereverberation in Numpy and Tensorflow for online and offline processing (2018, background): “p2 can be used but does not encourage further research and code changes [2].”
- Acoustic Modeling for Overlapping Speech Recognition: Jhu Chime-5 Challenge System (2019, methodology): “In this section, we investigate the effect of applying weighted prediction error (WPE) [26] based dereverberation before beamforming at test time.”
- Acoustic Modeling for Distant Multi-talker Speech Recognition with Single- and Multi-channel Branches (2019, methodology): “Then, we found that the 4-ch AM achieved a better, 63.16% WER without using any other SE techniques.1 By applying the WPE and NN-MVDR, we further ob- 1Note that the values in Table 2 are slightly better than the values reported tained about a 1.3-point absolute WER improvement, and we finally achiev”
- A Survey on Audio Diffusion Models: Text To Speech Synthesis and Enhancement in Generative AI (2023, unknown): “Specifically, the operator is initially estimated and adaptively corrected by weighted prediction error (WPE) [71].”
- Speech Dereverberation Using Non-Negative Convolutive Transfer Function and Spectro-Temporal Modeling (2016, methodology): “Probabilistic model-based methods [9]–[11] typically use an autoregressive process to model the reverberation, where the clean speech spectral coefficients are modeled using a Gaussian or a Laplacian distribution.”