Sortformer: A Novel Approach for Permutation-Resolved Speaker Supervision in Speech-to-Text Systems | Read Paper on Bytez