Software for recognizing speaker by voice

Abstract

. FBank (Filter Bank) is a front-end processing algorithm that processes audio in a way similar to the human ear and extracts features to improve the performance of speech recognition. The system uses an efficient context-aware masking-based network, CAM++, which uses a densely connected time-delay neural network (D-TDNN) as the backbone and adopts a novel multi-granularity pooling to capture different levels of context information.Based on the respective advantages of FBank and CAM++ models, this study designs a software for recognizing speaker by voice and implements the system through pytorch.