Hugging Face's logo

ad019el
/
Tamasheq-XLSR-53-mixed-english

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
wav2vec2
Inference Endpoints
Model card Files Metrics Community
6
ad019el commited on
Commit
3a1d4ee
•
1 Parent(s): b42ca29

add tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +45 -48
vocab.json CHANGED
@@ -1,50 +1,47 @@
1
  {
2
- "&": 5,
3
- "0": 28,
4
- "2": 0,
5
- "[PAD]": 47,
6
- "[UNK]": 46,
7
- "e": 25,
8
- "|": 12,
9
- "ء": 20,
10
- "آ": 23,
11
- "أ": 4,
12
- "ؤ": 37,
13
- "إ": 10,
14
- "ئ": 18,
15
- "ا": 16,
16
- "ب": 34,
17
- "ة": 3,
18
- "ت": 44,
19
- "ث": 35,
20
- "ج": 30,
21
- "ح": 9,
22
- "خ": 33,
23
- "د": 26,
24
- "ذ": 15,
25
- "ر": 11,
26
- "ز": 22,
27
- "س": 1,
28
- "ش": 31,
29
- "ص": 2,
30
- "ض": 7,
31
- "ط": 39,
32
- "ظ": 40,
33
- "ع": 13,
34
- "غ": 36,
35
- "ف": 38,
36
- "ق": 42,
37
- "ك": 45,
38
- "ل": 6,
39
- "م": 29,
40
- "ن": 27,
41
- "ه": 24,
42
- "و": 43,
43
- "ى": 21,
44
- "ي": 17,
45
- "ٱ": 14,
46
- "چ": 41,
47
- "ڤ": 32,
48
- "ک": 8,
49
- "ی": 19
50
  }
 
1
  {
2
+ "[PAD]": 44,
3
+ "[UNK]": 43,
4
+ "|": 10,
5
+ "ء": 18,
6
+ "آ": 21,
7
+ "أ": 3,
8
+ "ؤ": 33,
9
+ "إ": 8,
10
+ "ئ": 16,
11
+ "ا": 14,
12
+ "ب": 30,
13
+ "ة": 2,
14
+ "ت": 41,
15
+ "ث": 31,
16
+ "ج": 26,
17
+ "ح": 7,
18
+ "خ": 29,
19
+ "د": 23,
20
+ "ذ": 13,
21
+ "ر": 9,
22
+ "ز": 20,
23
+ "س": 0,
24
+ "ش": 27,
25
+ "ص": 1,
26
+ "ض": 5,
27
+ "ط": 35,
28
+ "ظ": 36,
29
+ "ع": 11,
30
+ "غ": 32,
31
+ "ف": 34,
32
+ "ق": 39,
33
+ "ك": 42,
34
+ "ل": 4,
35
+ "م": 25,
36
+ "ن": 24,
37
+ "ه": 22,
38
+ "و": 40,
39
+ "ى": 19,
40
+ "ي": 15,
41
+ "ٱ": 12,
42
+ "چ": 37,
43
+ "ڤ": 28,
44
+ "ک": 6,
45
+ "ی": 17,
46
+ "’": 38
 
 
 
47
  }