多轨道数字音频自适应变阶谱降噪模型构建  

Construction of adaptive variable order spectrum noise reductionmodel for multi⁃track digital audio

作　　者：文雅洁 陈娟[1] WEN Yajie;CHEN Juan(North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学,山西太原030051

出　　处：《现代电子技术》2023年第23期55-58,共4页Modern Electronics Technique

摘　　要：文中提出多轨道数字音频自适应变阶谱降噪模型构建,采用一阶高通数字滤波器预加重处理多轨道数字音频信号,以此为基础,通过最大熵谱估计算法估计数字音频信号频谱,搭建自适应变阶谱降噪模型,确定谱减阶数的自适应取值规则。将待处理的多轨道数字音频频谱估计结果输入至训练好的降噪模型中,输出结果经过逆变换即为降噪完成后的多轨道数字音频,从而实现了多轨道数字音频的自适应变阶谱降噪。实验数据显示:构建模型应用后,可以有效去除音频噪声信号,并不会缺失音频有效信息,降噪后多轨道数字音频信噪比最大值为91.25 dB,充分证实了构建模型降噪效果更佳。An adaptive variable order spectrum noise reduction model for multi⁃track digital audio is proposed.A first⁃order high⁃pass digital filter for pre emphasis processing of multi⁃track digital audio signals is adopted.On this basis,the maximum entropy spectral estimation algorithm is used to estimate the spectrum of the digital audio signal,an adaptive variable order spectrum noise reduction model is established,the adaptive value rules for the spectrum reduction order is determined,and the estimated results of the multi⁃track digital audio spectrum to be processed are input into the trained denoising model.The output result,after inverse transformation,will be the multi⁃track digital audio when noise reduction is fulfilled,which achieves adaptive variable order spectrum noise reduction for multi⁃track digital audio.The experimental data shows that the audio noise signals can be effectively removed without missing effective audio information after the application of the constructed model.The maximum value of multi⁃track digital audio signal⁃to⁃noise ratio(SNR)after noise reduction is 91.25 dB,which fully demonstrates that the constructed model has a better effect of noise reduction.

关 键 词：自适应变阶谱 阈值选取 降噪模型 多轨道数字音频 频谱分析 自适应谱估计 

分 类 号：TN911.4-34[电子电信—通信与信息系统] TP183[电子电信—信息与通信工程]