Voice interaction detection method, device, apparatus and storage medium

Abstract

Embodiments of the present application provide a voice interaction detection method and device, equipment and a storage medium, and relate to the technical field of intelligent devices. The voice signal corresponding to each voice collection unit of the voice to be detected is synchronously collected, and the device structure vibration information is synchronously collected by using an inertial measurement unit. The time-frequency transformation is performed on the voice signals of each channel to obtain the frequency domain representation of the voice signals. The spatial acoustic consistency feature is constructed based on the frequency domain representation, and the device structure vibration feature is constructed based on the device structure vibration information. The spatial acoustic consistency feature and the device structure vibration feature are fused to obtain a fusion feature vector. The pre-trained user voice feature model is used to calculate the score of the fusion feature vector to obtain a sentence-level score. The detection result of the voice to be detected is obtained according to the sentence-level score. The multi-modal voice liveness detection mechanism is realized, and the reliability and security of the voice interaction authentication system against complex fraud attacks are improved.

Images