Skip to main content
EURASIP Journal on Audio, Speech, and Music Processing

Table 3 Performance of Lhasa dialect ASR systems with four source languages as pretrained models

From: Improving low-resource Tibetan end-to-end ASR by multilingual and multilevel unit modeling

[JWFFXGRAPHICS]s13636-021-00233-4tmc3.eps[JWFFXGRAPHICS]

 
  1. The ASR model initialized with Aishell-1 train_sp significantly (the two-tailed t test at p value < 0.05) outperforms other models
Back to article page