Cong-Thanh Do, Mohan Li, Rama Doddipatla. Multiple-hypothesis RNN-T Loss for Unsupervised Fine-tuning and Self-training of Neural Transducer. In Hanseok Ko, John H. L. Hansen, editors, Interspeech 2022, 23rd Annual Conference of the International Speech Communication Association, Incheon, Korea, 18-22 September 2022. pages 4446-4450, ISCA, 2022. [doi]
Abstract is missing.