International Symposium on Chinese Spoken Language Processing (ISCSLP) is a well-known international conference in the field of speech and language processing. It is a flagship conference on Chinese speech processing in the world and is held every two years with a long history. Since 1998, 11 sessions have been successfully held. ISCSLP 2018 was held on November 26 - 29, 2018 in Taipei. It was jointly held by Academia Sinica and National Taiwan University, and was jointly supported by International Speech Communication Association (ISCA) and IEEE Signal Processing Society, and the papers were included by IEEE Xplore and indexed by EI.

The paper "Angular Softmax Loss for End-to-end Speaker Verification" of master student Li Yutian supervised by Associate Researcher Ou Zhijian from the Research Center for Media Big-data Cognitive Computing, Department of Electronic Engineering, Tsinghua University, won the Best Student Paper Award of this conference. This award selects 7 best paper candidates from 101 papers accepted by the conference and then selects only 2 papers to win this award through centralized defense.

Speaker recognition, also known as voiceprint recognition, is an important research area in the speech processing and AI field, with a wide and important application. This paper develops a new training criterion for the end-to-end speaker recognition system, and obtains an excellent recognition performance under the same experimental conditions, and is of general guiding significance for the pattern recognition research with category space as the open set.