A detection method and system for digital audio copy and paste tampering operation

Abstract

The invention belongs to the technical field of digital audio signal processing, and discloses a detection method and a system for digital audio copy and paste tampering operation. After removing silent segments of signals to be tested, the syllables are divided. Then, the syllable segments are pre-filled with zeros and post-filled with zeros, and the spectral map of each syllable segment is drawnand saved as a gray scale map. Through the scale-invariant feature transform (SIFT) for each generated spectral map, the feature point descriptors of the spectral map can be obtained, which can describe the syllable features. Based on similarity matching of feature point descriptors of two-syllable spectrogram, statistical rules are used to determine whether duplication occurs between two syllables. Paste Action. The invention improves the digital audio duplication. Paste detection accuracy while reducing the amount of computation, and can quickly complete the syllable matching process between the features, accurate positioning of the area with the occurrence of duplication-paste operation.

Description

technical field [0001] The invention belongs to the technical field of digital audio signal processing, and in particular relates to a detection method and system for digital audio copy and paste tampering operations. Background technique [0002] At present, the existing technologies commonly used in the industry are as follows: [0003] With the popularization of multimedia transmission equipment, voice signals are widely spread and applied in different fields. The convenience and entertainment of using voice signals bring a lot of fun to people's life, but there are also great potential safety hazards. Some criminals tamper with the voice signal, destroying the authenticity and integrity of the original voice signal, causing the original intention of the speaker to be distorted, the details covered, and even a forged voice to obtain benefits, access control rights, and obtain evidence in court Misleading the outcome of the trial. In recent years, digital audio forensic...

AI Technical Summary

Problems solved by technology

[0013] Endpoint detection and syllable segmentation are also important research topics in the field of digital signal processing. Effectively distinguishing voiced segments and syllable segmentation is critical to related fields of digital audio processing, but there is no widely recognized effective algorithm; audio signals and image signals Although they are all digital signals, some processing methods are very similar, but digital signals have their own uniqueness. When applying digital image related algorithms, attention should be paid to establishing an audio signal analysis model; digital audio copy-paste tampering detection involves meticulous and precise tampering Most of the syllable positioning requires a large amount of calculation, and it is difficult to determine the general decision-making conditions; the complexity and difficulty of the algorithm operation naturally lead to its low degree of automation, and it is difficult to have the significance of general promotion

Images