3 ms·Direct speech-to-speech translation with a sequence-to-sequence model1 points by lawrenceyan 7y ago