本文選題：聲紋比對 + 廣告智能監(jiān)播系統(tǒng)��； 參考：《電聲技術》2016年10期

【摘要】：主要通過比較目前主要的三種音頻檢測方式的優(yōu)缺點,依據(jù)電臺廣播與廣告的特點確定采用聲紋比對的解決方法,并制定了廣告智能監(jiān)播系統(tǒng)架構(gòu)。根據(jù)梅爾倒譜系數(shù)算法提取廣告原始音頻與錄播音頻、開路信號音頻特征值并進行對比的基本原理,確定了廣告智能監(jiān)播系統(tǒng)的初步流程與架構(gòu)。在實際測試中優(yōu)化了設計方案,最終測試結(jié)果達到既定目標,解決了北京電臺人工校驗廣告的問題,并為今后北京人民廣播電臺廣告智能監(jiān)播的進一步研發(fā)打好了基礎。
[Abstract]:By comparing the advantages and disadvantages of the three main audio detection methods at present, according to the characteristics of radio broadcasting and advertising, the solution of voice-pattern comparison is determined, and the framework of advertising intelligent monitoring and broadcasting system is developed. According to the principle of extracting the original audio frequency and recording audio frequency by Mel cepstrum coefficient algorithm and comparing the audio eigenvalues of open circuit signal, the preliminary flow and structure of the advertising intelligent monitoring and broadcasting system are determined. In the actual test, the design scheme is optimized, the final test result reaches the established goal, the problem of manual check advertisement of Beijing Radio Station is solved, and the foundation is laid for the further research and development of the intelligent supervision and broadcast of the advertisement of Beijing people's Radio Station in the future.
【作者單位】： 北京人民廣播電臺;
【分類號】：TN912.3;TN931.3
，

本文編號：1920437