Text-Independent Speaker Verification Based on Triplet Loss
Junjie He, Jing He, Liangjin Zhu · 2020 IEEE 4th Information Technology, Networking, Electronic and Automation Control Conference (ITNEC) · 2020
An improved end-to-end text-independent speaker verification model is proposed in this paper. LSTM networks are employed to extract the speaker model embedding, and the triplet loss is used to optimize the training of the network which make the training of the speaker verification model more efficient while keep the computation complexity relatively low. With the triplet loss, the proposed model can achieve better EER performance.