radiogroup-crits
commited on
Commit
•
b63099b
1
Parent(s):
63e7175
Update README.md
Browse files
README.md
CHANGED
@@ -1,78 +1,78 @@
|
|
1 |
-
---
|
2 |
-
language:
|
3 |
-
- it
|
4 |
-
license: apache-2.0
|
5 |
-
datasets:
|
6 |
-
- common_voice
|
7 |
-
- mozilla-foundation/common_voice_8_0
|
8 |
-
metrics:
|
9 |
-
- wer
|
10 |
-
- cer
|
11 |
-
tags:
|
12 |
-
- audio
|
13 |
-
- automatic-speech-recognition
|
14 |
-
- hf-asr-leaderboard
|
15 |
-
- it
|
16 |
-
- mozilla-foundation/common_voice_8_0
|
17 |
-
- speech
|
18 |
-
model-index:
|
19 |
-
- name: XLS-R Wav2Vec2 Italian by radiogroup crits
|
20 |
-
results:
|
21 |
-
- task:
|
22 |
-
name: Automatic Speech Recognition
|
23 |
-
type: automatic-speech-recognition
|
24 |
-
dataset:
|
25 |
-
name: Common Voice 8
|
26 |
-
type: mozilla-foundation/common_voice_8_0
|
27 |
-
args: it
|
28 |
-
metrics:
|
29 |
-
- name: Test WER
|
30 |
-
type: wer
|
31 |
-
value: 9.04
|
32 |
-
- name: Test CER
|
33 |
-
type: cer
|
34 |
-
value: 2.2
|
35 |
-
- name: Test WER (+LM)
|
36 |
-
type: wer
|
37 |
-
value: 6.24
|
38 |
-
- name: Test CER (+LM)
|
39 |
-
type: cer
|
40 |
-
value: 1.67
|
41 |
-
---
|
42 |
-
# XLS-R-1B-ITALIAN-DOC4LM-5GRAM
|
43 |
-
|
44 |
-
## Language model information
|
45 |
-
|
46 |
-
Our language model was generated using a dataset of Italian wikipedia articles and manual transcriptions about gr and television programs.
|
47 |
-
|
48 |
-
|
49 |
-
## Download CommonVoice8.0 dataset for italian language
|
50 |
-
```python
|
51 |
-
from datasets import load_dataset
|
52 |
-
|
53 |
-
dataset = load_dataset("mozilla-foundation/common_voice_8_0", "it", use_auth_token=True)
|
54 |
-
```
|
55 |
-
|
56 |
-
## Evaluation Commands
|
57 |
-
|
58 |
-
To evaluate on `mozilla-foundation/common_voice_8_0` with split `test`:
|
59 |
-
|
60 |
-
```bash
|
61 |
-
python eval.py --model_id radiogroup-crits/wav2vec2-xls-r-1b-italian-doc4lm-5gram --dataset mozilla-foundation/common_voice_8_0 --config it --split test --log_outputs --greedy
|
62 |
-
mv log_mozilla-foundation_common_voice_8_0_it_test_predictions.txt log_mozilla-foundation_common_voice_8_0_it_test_predictions_greedy.txt
|
63 |
-
mv mozilla-foundation_common_voice_8_0_it_test_eval_results.txt mozilla-foundation_common_voice_8_0_it_test_eval_results_greedy.txt
|
64 |
-
```
|
65 |
-
|
66 |
-
## Citation
|
67 |
-
If you want to cite this model you can use this:
|
68 |
-
|
69 |
-
```bibtex
|
70 |
-
@misc{radiogroup-crits2022wav2vec2-xls-r-1b-italian-doc4lm-5gram,
|
71 |
-
title={XLS-R Wav2Vec2 Italian by radiogroup-crits},
|
72 |
-
author={Raffaele Teraoni Prioletti
|
73 |
-
publisher={Hugging Face},
|
74 |
-
journal={Hugging Face Hub},
|
75 |
-
howpublished={\url{https://huggingface.co/radiogroup-crits/wav2vec2-xls-r-1b-italian-doc4lm-5gram}},
|
76 |
-
year={2022}
|
77 |
-
}
|
78 |
```
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- it
|
4 |
+
license: apache-2.0
|
5 |
+
datasets:
|
6 |
+
- common_voice
|
7 |
+
- mozilla-foundation/common_voice_8_0
|
8 |
+
metrics:
|
9 |
+
- wer
|
10 |
+
- cer
|
11 |
+
tags:
|
12 |
+
- audio
|
13 |
+
- automatic-speech-recognition
|
14 |
+
- hf-asr-leaderboard
|
15 |
+
- it
|
16 |
+
- mozilla-foundation/common_voice_8_0
|
17 |
+
- speech
|
18 |
+
model-index:
|
19 |
+
- name: XLS-R Wav2Vec2 Italian by radiogroup crits
|
20 |
+
results:
|
21 |
+
- task:
|
22 |
+
name: Automatic Speech Recognition
|
23 |
+
type: automatic-speech-recognition
|
24 |
+
dataset:
|
25 |
+
name: Common Voice 8
|
26 |
+
type: mozilla-foundation/common_voice_8_0
|
27 |
+
args: it
|
28 |
+
metrics:
|
29 |
+
- name: Test WER
|
30 |
+
type: wer
|
31 |
+
value: 9.04
|
32 |
+
- name: Test CER
|
33 |
+
type: cer
|
34 |
+
value: 2.2
|
35 |
+
- name: Test WER (+LM)
|
36 |
+
type: wer
|
37 |
+
value: 6.24
|
38 |
+
- name: Test CER (+LM)
|
39 |
+
type: cer
|
40 |
+
value: 1.67
|
41 |
+
---
|
42 |
+
# XLS-R-1B-ITALIAN-DOC4LM-5GRAM
|
43 |
+
|
44 |
+
## Language model information
|
45 |
+
|
46 |
+
Our language model was generated using a dataset of Italian wikipedia articles and manual transcriptions about gr and television programs.
|
47 |
+
|
48 |
+
|
49 |
+
## Download CommonVoice8.0 dataset for italian language
|
50 |
+
```python
|
51 |
+
from datasets import load_dataset
|
52 |
+
|
53 |
+
dataset = load_dataset("mozilla-foundation/common_voice_8_0", "it", use_auth_token=True)
|
54 |
+
```
|
55 |
+
|
56 |
+
## Evaluation Commands
|
57 |
+
|
58 |
+
To evaluate on `mozilla-foundation/common_voice_8_0` with split `test`:
|
59 |
+
|
60 |
+
```bash
|
61 |
+
python eval.py --model_id radiogroup-crits/wav2vec2-xls-r-1b-italian-doc4lm-5gram --dataset mozilla-foundation/common_voice_8_0 --config it --split test --log_outputs --greedy
|
62 |
+
mv log_mozilla-foundation_common_voice_8_0_it_test_predictions.txt log_mozilla-foundation_common_voice_8_0_it_test_predictions_greedy.txt
|
63 |
+
mv mozilla-foundation_common_voice_8_0_it_test_eval_results.txt mozilla-foundation_common_voice_8_0_it_test_eval_results_greedy.txt
|
64 |
+
```
|
65 |
+
|
66 |
+
## Citation
|
67 |
+
If you want to cite this model you can use this:
|
68 |
+
|
69 |
+
```bibtex
|
70 |
+
@misc{radiogroup-crits2022wav2vec2-xls-r-1b-italian-doc4lm-5gram,
|
71 |
+
title={XLS-R Wav2Vec2 Italian by radiogroup-crits},
|
72 |
+
author={Raffaele Teraoni Prioletti, Paolo Casagranda and Francesco Russo},
|
73 |
+
publisher={Hugging Face},
|
74 |
+
journal={Hugging Face Hub},
|
75 |
+
howpublished={\url{https://huggingface.co/radiogroup-crits/wav2vec2-xls-r-1b-italian-doc4lm-5gram}},
|
76 |
+
year={2022}
|
77 |
+
}
|
78 |
```
|