NARA-WPE: A Python package for weighted prediction error dereverberation in Numpy and Tensorflow for online and offline processing

134 citations recorded by the source corpus.

How researchers see this paper

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

CHiME-6 Challenge: Tackling Multispeaker Speech Recognition for Unsegmented Recordings (2020, methodology): “Both of them are combined with an open source version [41] of weighted prediction error (WPE) dereverberation [42], and integrated into our Kaldi recipe.”
A Review of Speaker Diarization: Recent Advances with Deep Learning (2021, methodology): “One of the most widely used techniques is the weighted prediction error (WPE) based dereverberation [89, 90, 91].”
The STC System for the CHiME-6 Challenge (2020, methodology): “DER reduction) on top of blockwise WPE dereverberation; • Fusion of probabilities from several TS-VAD models further improves diarization; • Best ASR results (up to 2.5% abs.”
Acoustic Modeling for Overlapping Speech Recognition: Jhu Chime-5 Challenge System (2019, methodology): “We explore the implementation of above mentioned techniques using publicly available tools including Nara WPE [12] and RIR generator [13].”
Front-end processing for the CHiME-5 dinner party scenario (2018, methodology): “The first algorithm applied to the microphone signals is multichannel WPE [16] for dereverberation, using our implementation [18] available on GitHub(1).”