Sign in
Compression of CTC-Trained Acoustic Models by Dynamic Frame-Wise Distillation or Segment-Wise N-Best Hypotheses Imitation.
Haisong Ding
Kai Chen
Qiang Huo
Published in:
INTERSPEECH (2019)
Keyphrases
</>
pairwise
neural network
speech recognition
data sets
learning algorithm
computer vision
multimedia
training set
semi supervised
probabilistic interpretation
discriminative training
acoustic models