khanhld
/

chunkformer-large-en-libri-960h

Automatic Speech Recognition

long-form transcription

Model card Files Files and versions

khanhld commited on Feb 10

Commit

9cee4fc

·

1 Parent(s): 4716077

[update] update model checkpoint, config, results

Files changed (3) hide show

README.md +1 -1
config.yaml +1 -1
pytorch_model.bin +2 -2

README.md CHANGED Viewed

@@ -74,7 +74,7 @@ We evaluate the models using **Word Error Rate (WER)**. To ensure a fair compari
 | STT | Model                 | Test-Clean | Test-Other | Avg.  |
 |-----|-----------------------|------------|------------|------ |
-|  1  | **ChunkFormer**       | 2.69       | 6.89       | 4.79  |
 |  2  | **Efficient Conformer** | 2.71     | 6.95       | 4.83  |
 |  3  | **Conformer**         | 2.77       | 6.93       | 4.85  |
 |  4  | **Squeezeformer**     | 2.87       | 7.16       | 5.02  |

 | STT | Model                 | Test-Clean | Test-Other | Avg.  |
 |-----|-----------------------|------------|------------|------ |
+|  1  | **ChunkFormer**       | 2.69       | 6.91       | 4.80  |
 |  2  | **Efficient Conformer** | 2.71     | 6.95       | 4.83  |
 |  3  | **Conformer**         | 2.77       | 6.93       | 4.85  |
 |  4  | **Squeezeformer**     | 2.87       | 7.16       | 5.02  |

config.yaml CHANGED Viewed

@@ -1,6 +1,6 @@
 encoder: chunkformer
 is_json_cmvn: true
-cmvn_file: chunkformer-large-en/global_cmvn
 input_dim: 80
 output_dim: 4990

 encoder: chunkformer
 is_json_cmvn: true
+cmvn_file: chunkformer-large-en-libri-960h/global_cmvn
 input_dim: 80
 output_dim: 4990

pytorch_model.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:015c879bbae04ef50b668bf6b54ae0777c9b461a2fa4db9ab485ef555d10cb04
-size 593420152

 version https://git-lfs.github.com/spec/v1
+oid sha256:568924c7472dfd7128f16088ef4fdb94296728a50ba7879a7fda8aea685c3ab3
+size 593426663