2 ms·Direct speech-to-speech translation with a sequence-to-sequence model2 points by sohkamyung 8y ago