search

syhw / wer_are_we

Attempt at tracking states of the arts and recent results (bibliography) on speech recognition.
1.86k stars 226 forks source link

Include "Conformer: Convolution-augmented Transformer for Speech Recognition" #46

Closed est31 closed 3 years ago

est31 commented 4 years ago

https://arxiv.org/abs/2005.08100

On the widely used LibriSpeech benchmark, our model achieves WER of 2.1%/4.3% without using a language model and 1.9%/3.9% with an external language model on test/testother. We also observe competitive performance of 2.7%/6.3% with a small model of only 10M parameters.

est31 commented 3 years ago

Done by #49 . Thanks!