LAS + SpecAugment

Reported on 2 benchmarks across 1 task · 1 paper · 2 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Audio2 results

Speech RecognitiononLibriSpeech test-clean
Word Error Rate (WER)· uses extra data· 2019-04-18
2.5
best: 0.985 (United Med ASR)
SOTA
SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition arXiv:1904.08779
Speech RecognitiononLibriSpeech test-other
Word Error Rate (WER)· uses extra data· 2019-04-18
5.8
best: 2.48 (SAMBA ASR)
SOTA
SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition arXiv:1904.08779