摘要
提出一种人工智能支持下的广电音频自动化降噪技术,旨在缓解广电音频中噪声干扰所引发的不利影响,优化用户体验。在对广电音频数据进行采集和预处理的基础上,借助梅尔频率倒谱系数(Mel-scale Frequency Cepstral Coefficients,MFCC)技术提取主要特征,而后基于卷积神经网络(Convolutional Neural Networks,CNN)模型实现对噪声的识别和去除。通过测试分析可知,所提方法在不同噪声环境中具有较为理想的降噪表现,能够有效提高广电音频质量。
The article proposes an automatic noise reduction technology for radio and television audio supported by artificial intelligence,aiming to alleviate the adverse effects caused by noise interference in radio and television audio and optimize the user experience.Based on the collection and preprocessing of radio and television audio data,the main features are extracted with the help of Mel-scale Frequency Cepstral Coefficients(MFCC)technology.Then,the recognition and removal of noise are achieved based on the Convolutional Neural Networks(CNN)model.Through test analysis,it can be known that the proposed method has a relatively ideal noise reduction performance in different noise environments and can effectively improve the audio quality of radio and television.
作者
黄铭
HUANG Ming(Chongzuo Preschool Education College,Chongzuo 532200,China)
出处
《电视技术》
2025年第6期128-131,共4页
Video Engineering
基金
教育部高等学校科学研究发展中心(2023IT107)。
关键词
广电音频
降噪
卷积神经网络(CNN)
radio and television audio
noise reduction
Convolutional Neural Network(CNN)
