电台广告智能监播系统设计  

Design of intelligent monitoring system for radio advertisement

作　　者：李怀畅[1] 

机构地区：[1]北京人民广播电台,北京100022

出　　处：《电声技术》2016年第10期20-25,32,共7页Audio Engineering

摘　　要：主要通过比较目前主要的三种音频检测方式的优缺点,依据电台广播与广告的特点确定采用声纹比对的解决方法,并制定了广告智能监播系统架构。根据梅尔倒谱系数算法提取广告原始音频与录播音频、开路信号音频特征值并进行对比的基本原理,确定了广告智能监播系统的初步流程与架构。在实际测试中优化了设计方案,最终测试结果达到既定目标,解决了北京电台人工校验广告的问题,并为今后北京人民广播电台广告智能监播的进一步研发打好了基础。In this paper, by comparing the advantages and disadvantages of three kinds of main audio detection, according to the characteristics of radio broadcasting and advertising determined by voiceprint comparison solution, and to develop the intelligent advertisement broadcast monitoring system architecture. According to basic principle of Mel Frequency Cepstrum Coefficient algorithm extracting advertisement for the original audio and recording audio, open audio signal eigenvalue and are compared to determine the intelligent advertisement broadcast monitoring system preliminary process and structure. In the actual test to optimize the design, the final test results achieve the established objectives to solve the Beijing People＇s Broad- casting Station artificial check advertising, and further research and development in the future the Beijing People＇S radio ad- vertising monitoring broadcast for lay a good foundation.

关 键 词：声纹比对 广告智能监播系统 MFCC 特征值提取 

分 类 号：TN931[电子电信—信号与信息处理]