Commit
·
b92cd0d
1
Parent(s):
d400ae7
adjust README
Browse files
README.md
CHANGED
|
@@ -78,9 +78,10 @@ The model can be evaluated as follows on the hungarian test data of Common Voice
|
|
| 78 |
|
| 79 |
```python
|
| 80 |
import torch
|
|
|
|
|
|
|
| 81 |
from datasets import load_dataset, load_metric
|
| 82 |
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
|
| 83 |
-
import re
|
| 84 |
|
| 85 |
LANG_ID = "hu"
|
| 86 |
MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-hungarian"
|
|
@@ -88,7 +89,7 @@ DEVICE = "cuda"
|
|
| 88 |
|
| 89 |
CHARS_TO_IGNORE = [",", "?", ".", "!", "-", ";", ":", '""', "%", "'", '"', "�", "ʿ", "·", "჻", "¿", "¡", "~", "՞",
|
| 90 |
"؟", "،", "।", "॥", "«", "»", "„", "“", "”", "「", "」", "‘", "’", "《", "》", "(", ")", "[", "]",
|
| 91 |
-
"=", "`", "_", "+", "<", ">", "…", "–", "°", "´", "ʾ"
|
| 92 |
|
| 93 |
test_dataset = load_dataset("common_voice", LANG_ID, split="test")
|
| 94 |
wer = load_metric("wer")
|
|
|
|
| 78 |
|
| 79 |
```python
|
| 80 |
import torch
|
| 81 |
+
import re
|
| 82 |
+
import librosa
|
| 83 |
from datasets import load_dataset, load_metric
|
| 84 |
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
|
|
|
|
| 85 |
|
| 86 |
LANG_ID = "hu"
|
| 87 |
MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-hungarian"
|
|
|
|
| 89 |
|
| 90 |
CHARS_TO_IGNORE = [",", "?", ".", "!", "-", ";", ":", '""', "%", "'", '"', "�", "ʿ", "·", "჻", "¿", "¡", "~", "՞",
|
| 91 |
"؟", "،", "।", "॥", "«", "»", "„", "“", "”", "「", "」", "‘", "’", "《", "》", "(", ")", "[", "]",
|
| 92 |
+
"=", "`", "_", "+", "<", ">", "…", "–", "°", "´", "ʾ"]
|
| 93 |
|
| 94 |
test_dataset = load_dataset("common_voice", LANG_ID, split="test")
|
| 95 |
wer = load_metric("wer")
|