An Adaptive Speech Homomorphic Encryption Scheme Based on Energy in Cloud Storage

Aiming at the problems of the traditional speech encryption scheme in cloud storage, such as security risks, excessive communication consumption, low robustness to resist multiple types of attacks, and low efficiency of the speech homomorphic encryption scheme, an adaptive speech homomorphic encryption scheme based on energy in cloud storage was proposed. Firstly, by comparing the threshold of speech energy, the improved Adaboost algorithm is used to design an adaptive classifier. Then, the speech data is divided into the sound and silent parts according to the energy threshold. Secondly, the BGV homomorphic encryption algorithm is used to encrypt the sound part of the information. Then, the Paillier homomorphic encryption algorithm is used to encrypt the silent part of the information. Finally, the two parts of ciphertext are combined to realize ciphertext domain operation and adaptive decryption. The experimental analysis shows that the proposed scheme has good encryption and decryption efficiency and low ciphertext expansion and can resist various attacks (including statistical, entropy, and chosen-plaintext attacks). © Institute of Mathematical Statistics, 2022