Enhanced-Deep-Residual-Shrinkage-Network-Based Voiceprint Recognition in the Electric Industry

Voiceprint recognition can extract voice features and identity the speaker through information, which has great application prospects in personnel verification dispatching electric industry. The traditional voiceprint algorithms work well a quiet environment. However, noise interference inevitably exists industry, degrading accuracy of algorithms. In this paper, we propose an enhanced deep residual shrinkage network (EDRSN)-based by combining with learning (DL) context noisy industry environment, where dual-path convolution recurrent (DPCRN) is employed to reduce noise, its structure also improved based on (DRSN). Moreover, further use convolutional block attention mechanism (CBAM) module hybrid dilated (HDC) proposed EDRSN. Simulation results show that enhance speaker’s vocal distinguish eliminate features, thus reducing influence achieving better performance