Hugging Face's logo

herwoww
/
mms-MGB3

Automatic Speech Recognition
Transformers
PyTorch
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Community
herwoww commited on
Commit
f1c722b
•
1 Parent(s): 0a684e5

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +88 -92
vocab.json CHANGED
@@ -1,98 +1,94 @@
1
  {
2
  "ara": {
3
  "#": 1,
4
- "'": 2,
5
- "(": 3,
6
- ")": 4,
7
- "-": 5,
8
- ".": 6,
9
- "A": 7,
10
- "C": 8,
11
- "D": 9,
12
- "E": 10,
13
- "F": 11,
14
- "G": 12,
15
- "H": 13,
16
- "I": 14,
17
- "L": 15,
18
- "M": 16,
19
- "N": 17,
20
- "O": 18,
21
- "P": 19,
22
- "R": 20,
23
- "S": 21,
24
- "T": 22,
25
- "U": 23,
26
- "V": 24,
27
- "Y": 25,
28
- "[PAD]": 93,
29
- "[UNK]": 92,
30
- "_": 26,
31
- "a": 27,
32
- "b": 28,
33
- "c": 29,
34
- "d": 30,
35
- "e": 31,
36
- "f": 32,
37
- "g": 33,
38
- "h": 34,
39
- "i": 35,
40
- "k": 36,
41
- "l": 37,
42
- "m": 38,
43
- "n": 39,
44
- "o": 40,
45
- "p": 41,
46
- "r": 42,
47
- "s": 43,
48
- "t": 44,
49
- "u": 45,
50
- "v": 46,
51
- "w": 47,
52
- "y": 48,
53
- "z": 49,
54
  "|": 0,
55
- "،": 50,
56
- "؟": 51,
57
- "ء": 52,
58
- "آ": 53,
59
- "أ": 54,
60
- "ؤ": 55,
61
- "إ": 56,
62
- "ئ": 57,
63
- "ا": 58,
64
- "ب": 59,
65
- "ة": 60,
66
- "ت": 61,
67
- "ث": 62,
68
- "ج": 63,
69
- "ح": 64,
70
- "خ": 65,
71
- "د": 66,
72
- "ذ": 67,
73
- "ر": 68,
74
- "ز": 69,
75
- "س": 70,
76
- "ش": 71,
77
- "ص": 72,
78
- "ض": 73,
79
- "ط": 74,
80
- "ظ": 75,
81
- "ع": 76,
82
- "غ": 77,
83
- "ـ": 78,
84
- "ف": 79,
85
- "ق": 80,
86
- "ك": 81,
87
- "ل": 82,
88
- "م": 83,
89
- "ن": 84,
90
- "ه": 85,
91
- "و": 86,
92
- "ى": 87,
93
- "ي": 88,
94
- "ً": 89,
95
- "ُ": 90,
96
- "ّ": 91
97
  }
98
  }
 
1
  {
2
  "ara": {
3
  "#": 1,
4
+ "-": 2,
5
+ ".": 3,
6
+ "A": 4,
7
+ "C": 5,
8
+ "D": 6,
9
+ "E": 7,
10
+ "F": 8,
11
+ "G": 9,
12
+ "H": 10,
13
+ "I": 11,
14
+ "L": 12,
15
+ "M": 13,
16
+ "N": 14,
17
+ "O": 15,
18
+ "P": 16,
19
+ "R": 17,
20
+ "S": 18,
21
+ "T": 19,
22
+ "U": 20,
23
+ "Y": 21,
24
+ "[PAD]": 89,
25
+ "[UNK]": 88,
26
+ "_": 22,
27
+ "a": 23,
28
+ "b": 24,
29
+ "c": 25,
30
+ "d": 26,
31
+ "e": 27,
32
+ "f": 28,
33
+ "g": 29,
34
+ "h": 30,
35
+ "i": 31,
36
+ "j": 32,
37
+ "k": 33,
38
+ "l": 34,
39
+ "m": 35,
40
+ "n": 36,
41
+ "o": 37,
42
+ "p": 38,
43
+ "r": 39,
44
+ "s": 40,
45
+ "t": 41,
46
+ "u": 42,
47
+ "v": 43,
48
+ "w": 44,
49
+ "y": 45,
50
+ "z": 46,
 
 
 
51
  "|": 0,
52
+ "،": 47,
53
+ "؟": 48,
54
+ "ء": 49,
55
+ "آ": 50,
56
+ "أ": 51,
57
+ "ؤ": 52,
58
+ "إ": 53,
59
+ "ئ": 54,
60
+ "ا": 55,
61
+ "ب": 56,
62
+ "ة": 57,
63
+ "ت": 58,
64
+ "ث": 59,
65
+ "ج": 60,
66
+ "ح": 61,
67
+ "خ": 62,
68
+ "د": 63,
69
+ "ذ": 64,
70
+ "ر": 65,
71
+ "ز": 66,
72
+ "س": 67,
73
+ "ش": 68,
74
+ "ص": 69,
75
+ "ض": 70,
76
+ "ط": 71,
77
+ "ظ": 72,
78
+ "ع": 73,
79
+ "غ": 74,
80
+ "ف": 75,
81
+ "ق": 76,
82
+ "ك": 77,
83
+ "ل": 78,
84
+ "م": 79,
85
+ "ن": 80,
86
+ "ه": 81,
87
+ "و": 82,
88
+ "ى": 83,
89
+ "ي": 84,
90
+ "ً": 85,
91
+ "ُ": 86,
92
+ "ّ": 87
 
93
  }
94
  }