Fetching the paper…

RWTH ASR Systems for LibriSpeech: Hybrid vs Attention -- w/o Data Augmentation · Around