Detection of different voice diseases based on the nonlinear characterization of speech signals

作者：

Highlights：

• A novel methodology to characterize voice diseases using nonlinear dynamics.

• Use of complexity measures based on the analysis of the time delay embedded space.

• Transformation of the feature space using a Discrete Hidden Markov Model.

• The methodology validated on three different datasets with different voice diseases.

摘要

•A novel methodology to characterize voice diseases using nonlinear dynamics.•Use of complexity measures based on the analysis of the time delay embedded space.•Transformation of the feature space using a Discrete Hidden Markov Model.•The methodology validated on three different datasets with different voice diseases.

论文关键词：Nonlinear dynamic parameterization,Hidden Markov models,Laryngeal pathologies,Hypernasality,Disarhtria,Disphonia,Parkinson's disease,Speech signal

论文评审过程：Received 13 September 2016, Revised 30 March 2017, Accepted 4 April 2017, Available online 8 April 2017, Version of Record 19 April 2017.

论文官网地址：https://doi.org/10.1016/j.eswa.2017.04.012