Year: 2024
Type: conference-paper
Study design: Clinical Trial
Abstract: This study aims to explore the noise reduction performance of an AI-powered audio noise reduction system and its practical application in remote education. Addressing limitations of traditional audio noise reduction systems, such as operational complexity and high hardware requirements, this research tries to improve these by introducing an AI-powered noise reduction system for remote teaching audio. Fourteen teaching audio recordings were processed with the system, and objective metrics such as Signal-to-Noise Ratio (SNR) and Signal Distortion Ratio (SDR) were used to assess the denoise effectiveness of the system. Furthermore, a remote education experiment was designed to assess the subjective audio quality and psychological impact of the teaching audio recordings. Participants were randomly allocated to either audio learning group or waiting-list group. Results show significant enhancements both for objective metrics and subjective evaluations of noise-reduced audio. The intervention group gave a high rating regarding the audio quality and had significant improvements in self-esteem levels. These findings prove the superior noise reduction capability of the system and provide a technical reference for educators in developing teaching audio while supporting the advancement of remote education.
Authors: Xinlei Feng, Yuzheng Wang, Fei Luo
Language: English
Topic: Speech and Audio Processing
Subfield: Signal Processing
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: bronze
OpenAlex ID: W4410210747