Table 1 Duration information of three open source datasets.
From: An Mcformer encoder integrating Mamba and Cgmlp for improved acoustic feature extraction
Dataset | Language | Duration (h) | ||
|---|---|---|---|---|
Train | Dev | Test | ||
Aishell-1 | Mandarin | 150.85 | 18.09 | 10.03 |
CommonVoice zh 14 | Mandarin | 214.47 | 15.92 | 17.45 |
TED-LIUM 3 | English | 452.06 | 1.60 | 2.62 |
