A speaker recognition algorithm based on densely connected time delay neural network
Speaker recognition is an important biometric identification technology.In recent years,speaker recognition algorithms that use time delay neural network to extract vocal features have achieved outstanding results.To further enhance the ability of time delay neural network to extract speaker features and improve the recognition accuracy without consuming too much computational resources,a densely connected time delay neural network with an attention mechanism is proposed for speaker recognition by investigating existing speaker recognition algorithms.The densely connected structure enhances the information reuse between different network layers while effectively controlling the model size.The channel attention mechanism and frame attention mechanism help the network to focus on more critical details of the features,making the speaker features extracted by statistical pooling more representative.Experimental results show that an equal error rate(EER)of 1.40％and a minimum detection cost criterion(MinDCF)of 0.15 were achieved on the VoxCeleb1 test dataset,demonstrating effectiveness on the speaker recognition task.
万方数据
维普
