SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing Paper • 2110.07205 • Published Oct 14, 2021 • 5 • 5