An on-line VAD based on Multi-Normalisation Scoring (MNS) of observation likelihoods

作者：

Highlights：

• A novel on-line voice activity detection (VAD) technique with no delay is proposed.

• The VAD relies on Multi-Normalization Scoring (MNS) and a classifier.

• MNS is based on generating multiple score vectors with various normalization ratios.

• A validation experiment with two standard VADs proves its good performance.

摘要

•A novel on-line voice activity detection (VAD) technique with no delay is proposed.•The VAD relies on Multi-Normalization Scoring (MNS) and a classifier.•MNS is based on generating multiple score vectors with various normalization ratios.•A validation experiment with two standard VADs proves its good performance.

论文关键词：VAD,Observation likelihood,MNS,On-line speech processing

论文评审过程：Received 27 December 2017, Revised 25 May 2018, Accepted 30 May 2018, Available online 31 May 2018, Version of Record 5 June 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.05.038