透過您的圖書館登入
IP:18.226.98.183
  • 期刊

Multi-format Speech Perception Hashing Algorithm Based on Short-Time Logarithmic Energy and Improved Mel Energy Parameter Fusions

摘要


Aiming at the problems of single speech format, non- universal algorithm and low accuracy of tamper detection and location in existing speech content authentication algorithms, a multi-format speech perception hashing algorithm based on short-time logarithmic energy and improved energy parameter fusion is proposed. Firstly, the speech signal to be processed is preprocessed, and the short-time logarithmic energy and improved Mel energy of each frame are calculated. Then, perform time- frequency parameter fusion on time-frequency features by mean filtering, and the time-frequency parameters are constructed by difference hashing method. Finally, in order to improve the security of the algorithm, logical chaotic map is used to encrypt hash sequences with equal length scrambling. The experimental results show that the proposed algorithm not only has a good compromise between robustness and discrimination, but also has good robustness, discrimination and key dependence for multi- format speech signals, and can achieve small-scale tamper detection and location.

延伸閱讀