Sound quality prediction and improving of vehicle interior noise based on deep convolutional neural networks

作者：

Highlights：

• Deep CNNs with DropConnect are proposed to predict vehicle interior sound quality.

• The time and frequency characteristics of noise are simultaneously considered.

• Deep CNNs with time–frequency images as input perform better than other methods.

• The neuron visualization algorithm is introduced for feature visualization and understanding.

摘要

•Deep CNNs with DropConnect are proposed to predict vehicle interior sound quality.•The time and frequency characteristics of noise are simultaneously considered.•Deep CNNs with time–frequency images as input perform better than other methods.•The neuron visualization algorithm is introduced for feature visualization and understanding.

论文关键词：Vehicle interior noise,Sound quality,Subjective evaluation,Convolutional neural networks,Feature visualization

论文评审过程：Received 10 November 2019, Revised 28 May 2020, Accepted 11 June 2020, Available online 25 June 2020, Version of Record 21 July 2020.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.113657