Comparison of Model Performance Under Different Data Augmentation Strategies on the ST-CMDS-PHONEME Dataset.

Comparison of Model Performance Under Different Data Augmentation Strategies on the ST-CMDS-PHONEME Dataset.

<p>Comparison of Model Performance Under Different Data Augmentation Strategies on the ST-CMDS-PHONEME Dataset.</p>

Saved in:

Bibliographic Details
Main Author:	Zhaohui Du (414105) (author)
Other Authors:	Xiaofeng Zhao (413698) (author), Lin Li (28817) (author), Baohua Yu (733656) (author), Lijiang Miao (21416712) (author)
Published:	2025
Subjects:	Science Policy Space Science Biological Sciences not elsewhere classified Information Systems not elsewhere classified word error rate stateless pred network reduce parameter count phonemes recognition method method performs excellently improve recognition accuracy gelu activation function experimental results demonstrate determining mispronounced phonemes artificial intelligence technologies mandarin pronunciation based phoneme recognition task mandarin phoneme recognition optimizing recognition performance modeling </ p zipformer block architecture hybrid pruned rnn pronunciation evaluation zipformer encoder improved zipformer phoneme sequences phoneme datasets phoneme dataset xlink "> three layers sequence comparison recent years paper proposes model requires hot topic gradually become data preprocessing converting speech 1m parameters
Tags:	Add Tag No Tags, Be the first to tag this record!

Be the first to leave a comment!