【摘要】：隨著科技的進(jìn)步和發(fā)展，廣播電視技術(shù)也在不斷進(jìn)步，廣播電視節(jié)目的數(shù)量也增加很多。傳統(tǒng)的監(jiān)測方式受到極大的挑戰(zhàn)，因此需要新型自動監(jiān)測系統(tǒng)出現(xiàn)來適應(yīng)廣播電視技術(shù)的發(fā)展。同時現(xiàn)在市面上在使用的音頻監(jiān)測系統(tǒng)主要是基于文本庫的監(jiān)測比對方式，，這種監(jiān)測方式受限于文本庫的大小和內(nèi)容，不能普遍使用。因此，基于內(nèi)容的音頻監(jiān)測技術(shù)隨之得到廣泛的研究。這要求不同于以前文本庫的方式，需要提取能反映音頻信號特點的特征參數(shù)信息，使得系統(tǒng)能有更強(qiáng)的適應(yīng)能力。同時基于內(nèi)容的音頻監(jiān)測技術(shù)是使用實時監(jiān)測的方法，能夠快速給出監(jiān)測結(jié)果。 本文研究的是基于內(nèi)容的音頻監(jiān)測問題。研究的目的是監(jiān)測是否有播放內(nèi)容的變化，主要涉及播出節(jié)目和廣告兩個方面。這樣既保護(hù)了廣播電視制作方的知識產(chǎn)權(quán)，同時也讓老百姓能看到安全放心的廣播電視節(jié)目。系統(tǒng)使用了FPGA芯片，使得最后做出的系統(tǒng)體積小，而且便于系統(tǒng)更新升級。 本文根據(jù)監(jiān)測系統(tǒng)的要求，詳細(xì)的分析了語音信號的主要特征參數(shù)，主要介紹了語音信號的時域特征參數(shù)，最終根據(jù)實際使用情況選擇了短時平均過零率和短時平均幅度電平兩個特征參數(shù)共同作為監(jiān)測系統(tǒng)的特征參數(shù)。其中這兩個特征參數(shù)都是基于語音信號的短時特性，這就要求在特征數(shù)據(jù)提取之前必須要先對語音信號做加窗分幀預(yù)處理�？紤]到作為電視伴音監(jiān)測系統(tǒng)，于是選擇了視頻信號中的ODD信號作為分幀信號，這樣能使監(jiān)測比對的兩路信號做到粗同步的要求。這兩個特征參數(shù)的數(shù)據(jù)量也很小，便于遠(yuǎn)端提取出來的特征數(shù)據(jù)通過網(wǎng)絡(luò)傳輸?shù)奖镜乇O(jiān)測端。在經(jīng)過很多不同的傳輸路徑之后，兩路信號在時間上已經(jīng)不同步了，所以信號的同步對齊是該系統(tǒng)需要重要解決的問題�；趯ρ舆t信息的分析，提出了搜索跟蹤技術(shù)來解決信號的對齊問題，即在本地端加固定整數(shù)幀和小數(shù)幀延遲去逼近遠(yuǎn)端的特征數(shù)據(jù)。為了加快搜索跟蹤的速度，使用了照射體制，即使用連續(xù)多幀本地特征數(shù)據(jù)跟遠(yuǎn)端特征數(shù)據(jù)同時進(jìn)行比較。最后對系統(tǒng)的各個模塊做了仿真驗證。
[Abstract]:With the progress and development of science and technology, radio and television technology is also progressing, the number of radio and television programs is also increasing a lot. Traditional monitoring methods are greatly challenged, so a new automatic monitoring system is needed to adapt to the development of radio and television technology. At the same time, the audio monitoring system in use on the market is mainly based on the text library of the monitoring and comparison mode, this monitoring mode is limited by the size and content of the text library, can not be widely used. Therefore, content-based audio monitoring technology has been widely studied. This requires that the feature parameter information which can reflect the characteristics of audio signal should be extracted, which is different from that of the previous text library, so that the system can have stronger adaptability. At the same time, the content-based audio monitoring technology is a real-time monitoring method, which can give the monitoring results quickly. In this paper, the content-based audio monitoring problem is studied. The purpose of the study is to monitor whether there is a change in broadcast content, mainly involving broadcast programs and advertising. This not only protects the intellectual property rights of radio and television producers, but also allows ordinary people to see safe and secure radio and television programs. The FPGA chip is used in the system, which makes the final system small and easy to update and upgrade. According to the requirement of monitoring system, the main characteristic parameters of speech signal are analyzed in detail, and the time domain characteristic parameters of speech signal are mainly introduced. Finally, two characteristic parameters, short-time average zero-crossing rate and short-time average amplitude level, are selected as the characteristic parameters of the monitoring system. These two feature parameters are based on the short-term characteristics of the speech signal, which requires that the speech signal must be pre-processed by windowing and framing before the feature data is extracted. Considering that it is a TV sound monitoring system, the ODD signal in the video signal is chosen as the framing signal, so that the two signals of the monitoring comparison can be coarsely synchronized. The data amount of these two feature parameters is also very small, so it is convenient to transfer the feature data extracted from the remote end to the local monitoring end through the network. After many different transmission paths, the two signals are out of synchronization in time, so the synchronization of the signals is an important problem to be solved in the system. Based on the analysis of delay information, a search-tracking technique is proposed to solve the problem of signal alignment, in which fixed integer frames and decimal frames are added to the local end to approximate the characteristic data of the remote end. In order to accelerate the speed of searching and tracking, an irradiating system is used, even if the continuous multi-frame local feature data is compared with the remote feature data at the same time. Finally, each module of the system is simulated and verified.
【學(xué)位授予單位】：電子科技大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2013
【分類號】：TN941.1

【參考文獻(xiàn)】

相關(guān)期刊論文 前9條

1 郭利剛;方土富;;淺析廣播電視廣告節(jié)目監(jiān)測技術(shù)[J];中國傳媒科技;2007年01期

2 梁達(dá)先;;一種射頻視音頻中斷監(jiān)測器[J];電視技術(shù);2008年06期

3 崔文沖;;應(yīng)用于廣播電視音頻監(jiān)測系統(tǒng)的技術(shù)分析[J];電視技術(shù);2011年22期

4 李濤;電視廣告計算機(jī)監(jiān)審管理系統(tǒng)的研制與應(yīng)用[J];廣播與電視技術(shù);2005年01期

5 陳月昌;肖蓉;武愛敏;;數(shù)字音頻監(jiān)測系統(tǒng)設(shè)計及應(yīng)用[J];廣播與電視技術(shù);2007年05期

6 崔冬青,李治柱,吳亞棟;一種噪聲環(huán)境下連續(xù)語音識別的快速端點檢測算法[J];計算機(jī)工程與應(yīng)用;2003年23期

7 黃知義,周寧;基于內(nèi)容視頻檢索的關(guān)鍵技術(shù)研究[J];現(xiàn)代情報;2005年10期

8 李國輝,李恒峰;基于內(nèi)容的音頻檢索:概念和方法[J];小型微型計算機(jī)系統(tǒng);2000年11期

9 馬彬;淺談有線電視播出信號實時監(jiān)測系統(tǒng)的運(yùn)用[J];有線電視技術(shù);2005年18期

相關(guān)碩士學(xué)位論文 前1條

1 周慶華;基于嵌入式Linux說話人識別系統(tǒng)的研究[D];西華大學(xué);2009年

本文編號：2449648