项目名称: 数字音频被动取证关键技术研究
项目编号: No.61501199
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 无线电电子学、电信技术
项目作者: 王志锋
作者单位: 华中师范大学
项目金额: 19万元
中文摘要: 项目针对数字音频数据面临的易被篡改、伪造、盗用等安全问题,研究数字音频被动取证关键技术,通过直接分析数字音频信号本身,利用开集录音设备源识别算法对数字音频的原始性进行验证,利用数字音频篡改盲检测和定位算法对数字音频的真实性进行检测。项目主要研究内容包括:(1)针对开集录音设备源识别中设备特征的表达问题,研究语音段与非语音段中设备特征的提取与融合方法;(2)针对开集录音设备源识别中设备模型的训练问题,研究集内、集外设备模型的建模与模型参数优化方法,以及集内、集外设备最优判决阈值的训练方法;(3)针对数字音频中插入、删除、替换和拼接等篡改操作,研究相应的篡改盲检测方法,以及精确的篡改点定位方法。项目从设备信息与数字音频的特有关联性出发,利用现代信号处理和机器学习的手段,研究“录音设备指纹”的表征与建模方法,为数字音频被动取证技术的进一步研究提供新思路和理论依据,该研究成果在司法取证、知识产权保护等领域有广泛的市场化应用前景。
中文关键词: 多媒体取证;语音信息处理;数字音频取证;开集设备源识别;篡改点检测
英文摘要: Aimed to deal with several security issues such as tampering, forgery, data theft faced by digital audio, this project focuses on the key technologies of digital audio forensics, which directly analysis of the digital audio itself to verify the integrity through open-set source recording devices identification and the authenticity through blind tampering detection. The main research contents of this project include: (1) For the feature expressions in open-set source recording devices identification, this project plans to do research work on feature fusion to combine the device features both in speech and non-speech periods;(2) For the modeling problems in open-set source recording devices identification, this project plans to do research work on in-set/out-set device modeling, model optimization and the training method of the best threshold;(3)For detecting the tampering such as inserting, deleting, replacing and splicing in digital audio data, this project plans to do research work on blind tampering detection and tampering points localization. This project considers about the relevance between devices and digital audio, and it uses modern signal processing and machine learning methods to present “recording device fingerprint”, which are for the purpose of further research and application of digital audio forensics technology and new ideas and theoretical basis. This research can be applied in Count Forensics, the protection of intellectual property rights.
英文关键词: Multimedia Forensics;Speech Information Processing;Digital Audio Forensics;Open-set Source Recording Devices Identification;Detection of Tampering Points