基于MFCC的空中交通管制语音指令的特征提取研究  

The Research on Feature Extraction of Air Traffic Control Voice Instructions Based on MFCC

作　　者：王兴林 WANG Xinglin(Yunnan Sub-Bureau of Southwest Regional Air Traffic Management Bureau CAAC,Kunming 650000,China)

机构地区：[1]中国民用航空西南地区空中交通管理局云南分局,云南昆明650000

出　　处：《电声技术》2023年第6期68-72,共5页Audio Engineering

摘　　要：随着人工智能技术的不断应用,智能安全与智慧民航不断深入发展,通过科技手段提高空中交通飞行安全成为全球民航的共同选择。空中交通管制语音数据作为民航新型生产要素,对语音信号进行深入研究并合理运用,对于提高飞行安全具有重要意义。文章介绍空中交通管制指令的基本要求,详细分析语音信号特征提取的各个环节,通过梅尔频率倒谱系数(Mel Frequency Cepstrum Coefficient,MFCC)提取语音信号的特征,再使用高斯混合模型(Gaussian Mixture Model,GMM)进行训练和分类,从而实现语音信号的识别,具有一定的实际运用价值。With the continuous application of artificial intelligence technology,intelligent safety and smart civil aviation are developing in depth,and improving air traffic flight safety through scientific and technological means has become the common choice of civil aviation all over the world.As a new production factor of civil aviation,air traffic control voice data is of great significance to improve flight safety by in-depth study and rational application of voice signals.This paper introduces the basic requirements of air traffic control instructions,and analyzes in detail all aspects of voice signal feature extraction.The features of voice signal are extracted by Mel Frequency Cepstrum Coefficients(MFCC),and then trained and classified by Gaussian Mixture Model(GMM),so as to realize voice signal recognition,which has certain practical application value.

关 键 词：语音信号 特征提取 梅尔频率倒谱系数(MFCC) 高斯混合模型(GMM) 模型训练 

分 类 号：TN912.34[电子电信—通信与信息系统]