Method for mobile phone source identification in additive noise environment based on constant Q transform domain

Abstract

The invention discloses a method for mobile phone source identification an additive noise environment based on the constant Q transform domain, which is characterized in that a spectrum distribution characteristic vector of the constant Q transform domain is obtained by using constant Q transform, wherein the constant Q transform has higher frequency resolution at a low frequency and has higher time resolution at a high frequency, thereby being more suitable for mobile phone source identification; a multi-scene training mode is adopted in the training stage, there are not only clean voice samples without scene noise but also noisy voice samples with different scene noise types and noise intensities in a training set, an M-classification model obtained by the training is universal and can carries out effective mobile phone source identification on voice samples with known noise scenes and unknown noise scenes; and a CNN model of deep learning is used to establish the M-classification model, the CNN model not only improves the accuracy of source identification for the clean voice samples without scene noise but also greatly improves the effect of mobile phone source identification for the noisy voice samples, and the noise robustness is strong.

Description

technical field [0001] The invention relates to a mobile phone source identification technology, in particular to a mobile phone source identification method in an additive noise environment based on a constant Q transform domain. Background technique [0002] With the continuous development and progress of digital multimedia and Internet technology, a variety of powerful and easy-to-operate digital media editing software has emerged, which brings new problems and challenges to the availability of collected data—multimedia security issues. As a technology to detect the originality, authenticity and integrity of multimedia data, multimedia forensics technology is a hot research issue in the field of information security. As a branch of multimedia forensics technology, the source identification of recording equipment has great research significance. Compared with recording pens, cameras, DV and other equipment, mobile phones are more popular and convenient. More and more peopl...

AI Technical Summary

Problems solved by technology

[0006]Although most mobile phone recording device recognition algorithms have good accuracy in device recognition, there are still some limitations, and few studies will consider noise attacks The robustness of the algorithm in the case of

However, in real life, the recording files that need to be identified are usually recorded in different noise environments, and the source identification of mobile phone recordings in noisy environments is more realistic and challenging.

Images