Research On AAC Audio Transcoding Detection

The development of multimedia technology has brought many conveniences to people's lives,but the authenticity and reliability of its information have become increasingly prominent.As a result,digital multimedia forensics emerged as the times require and developed rapidly.Audio forensics has also evolved.Identifying the authenticity and reliability of audio content has become one of the hot issues in current research.Digital audio transcoding detection is one of the important components of audio forensics.There are two forms of digital audio transcoding.One is transcoding between audio code rates in the same format,i.e.,double compression of audio,and the other is transcoding between audio formats in different formats.As the current popular AAC audio,it is the focus of this research.This paper uses the change of codec parameters from the quantization step.It mainly studies the detection algorithm of AAC audio transcoding and explores the detection of audio transcoding in the same format and the detection of audio transcoding in different formats.In this paper,three aspects are mainly studied.Firstly,the AAC codec principle is deeply analyzed,and the important components of the AAC coding process are introduced in detail,which provides a theoretical basis for the subsequent research work.Based on this,we establish a basic sample database of audio transcoding in the same format and different formats.Secondly,through research,it is found that as the number of AAC audio compression increases,the scale factor will also change.We use the difference of eigenvalue changes after AAC audio is compressed again,and propose two AAC transcoding audio detection algorithms based on scale factor.The first is the AAC dual-compressed audio detection algorithm based on the scale factor difference.The experimental results show that the proposed algorithm has better detection performance than the existing algorithm.The accuracy of AAC audio detection between low bit rate transcoding to high code rate is 99.91%,and the detection accuracy of the same rate audio compression is also 97.98%.The experimental results of different encoders,different durations,and different noise interferences also prove the effectiveness of the algorithm.The second is the AAC audio compression history detection algorithm based on the scale factor transition probability.The experimental results show that the proposed algorithm can effectively detect the AAC audio compression history.The accuracy of AAC audio detection between low bit rate transcoding to high code rate is 96.7%,and the detection accuracy between the same code rate is also 92.16%.The comparison experiment also proves that the algorithm has better performance.Finally,it is found through research that AAC audio will leave a certain quantization trace after compression,and the quantized trace will be erased to some extent after removing some sampling points.Based on the theoretical analysis of the quantized traces,we propose two kinds of AAC transcoded audio detection algorithms based on QMDCT coefficients.One is the AAC dual compression audio detection algorithm based on QMDCT coefficients.The experimental results show that the proposed algorithm can effectively detect AAC transcoded audio,and the accuracy of AAC audio detection between low bit rate transcoding to high code rate is 99.84%,and the detection accuracy between the same code rate is also 98.60%.The second is an audio transcoding detection algorithm based on different formats of QMDCT coefficients.The detection of the audio transcoding history and the detection of the original audio format are performed separately.The experimental results show that the algorithm can effectively detect whether AAC audio has been transcoded,and its average detection accuracy is 99.28%.And the exact classification of which transcoding is carried out,the average detection accuracy rate is 97.68%.