Huang S.-FChuang S.-PLiu D.-RChen Y.-CYang G.-PHUNG-YI LEE2022-04-252022-04-2520212308457Xhttps://www.scopus.com/inward/record.uri?eid=2-s2.0-85119205277&doi=10.21437%2fInterspeech.2021-763&partnerID=40&md5=ec4477ece29732cedbc0c143ad1eb2dahttps://scholars.lib.ntu.edu.tw/handle/123456789/607149Speech separation has been well developed, with the very successful permutation invariant training (PIT) approach, although the frequent label assignment switching happening during PIT training remains to be a problem when better convergence speed and achievable performance are desired. In this paper, we propose to perform self-supervised pre-training to stabilize the label assignment in training the speech separation model. Experiments over several types of self-supervised approaches, several typical speech separation models and two different datasets showed that very good improvements are achievable if a proper self-supervised approach is chosen. Copyright ? 2021 ISCA.Label Permutation SwitchSelf-supervised Pre-trainSpeech EnhancementSpeech SeparationSeparationSource separationSpeech analysisSpeech communicationAchievable performanceConvergence speedLabel permutation switchPre-trainingSelf-supervised pre-trainSeparation modelSpeech separationSpeed performanceSpeech enhancementconference paper10.21437/Interspeech.2021-7632-s2.0-85119205277