SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition

Benchmark Model Rank Results
speech-recognition-on-librispeech-test-cleanLAS + SpecAugment#92Word Error Rate (WER): 2.5
speech-recognition-on-librispeech-test-cleanLAS (no LM)#95Word Error Rate (WER): 2.7
speech-recognition-on-librispeech-test-otherLAS + SpecAugment#90Word Error Rate (WER): 5.8
speech-recognition-on-librispeech-test-otherLAS (no LM)#96Word Error Rate (WER): 6.5