基于子带处理与Volterra自适应滤波的广播音频信号相似性检测方法

Broadcast Audio Signal Similarity Detection Method Based on Subband Processing and Volterra Adaptive Filtering

由于广播节目受众广,影响力大,其播控要求高,对错播、插播、漏播等异常播出情况容忍度低。针对广播节目播出实时监测问题,本文提出了一种快速的广播音频信号相似性检测方法。该方法计算Pearson相关系数来判别两广播音频信号是否相似。然后,为了抵消编解码器、收发设备及传输信道的影响,应用自适应Volterra滤波器来处理信号。最后,用子带分解技术将全频带信号分解为子带信号,并仅对功率最高的子带进行分析预处理,以降低计算量。实验结果表明,在考虑了真实的传输影响后,通过不同条件的比较,该方法具有良好的检测准确度,且计算量较小,可以满足实时处理的要求。

Due to the wide audience and great influence of radio programs,its broadcast control requirements were high,and the tolerance for abnormal broadcasts such as mis-broadcasting,interrupted broadcasts,and missed broadcasts was low.Aiming at the real-time monitoring of broadcast programs,this paper proposed a fast broadcast audio signal similarity detection method.This method calculated the Pearson correlation coefficient to determine whether the two broadcast audio signals were similar.Then,in order to offset the influence of the codec,transceiver equipment and transmission channel,an adaptive Volterra filter was applied to process the signal.Finally,the sub-band decomposition technology was used to decompose the full-band signal into sub-band signals,and only the sub-band with the highest power was analyzed and pre-processed to reduce the amount of calculation.The experimental results show that,after considering the real transmission impact and comparing different conditions,the method has good detection accuracy and a small amount of calculation,which can meet the requirements of real-time processing.