Hugging Face's logo

Thienpkae
/
wav2vec2-vivos-asr

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
wav2vec2
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
Thienpkae commited on
Commit
f6bd907
•
1 Parent(s): 6e5d305

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +89 -89
vocab.json CHANGED
@@ -1,94 +1,94 @@
1
  {
2
- "A": 1,
3
- "B": 2,
4
- "C": 3,
5
- "D": 4,
6
- "E": 5,
7
- "G": 6,
8
- "H": 7,
9
- "I": 8,
10
- "K": 9,
11
- "L": 10,
12
- "M": 11,
13
- "N": 12,
14
- "O": 13,
15
- "P": 14,
16
- "Q": 15,
17
- "R": 16,
18
- "S": 17,
19
- "T": 18,
20
- "U": 19,
21
- "V": 20,
22
- "X": 21,
23
- "Y": 22,
24
  "[PAD]": 91,
25
  "[UNK]": 90,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
26
  "|": 0,
27
- "À": 23,
28
- "Á": 24,
29
- "Â": 25,
30
- "Ã": 26,
31
- "È": 27,
32
- "É": 28,
33
- "Ê": 29,
34
- "Ì": 30,
35
- "Í": 31,
36
- "Ò": 32,
37
- "Ó": 33,
38
- "Ô": 34,
39
- "Õ": 35,
40
- "Ù": 36,
41
- "Ú": 37,
42
- "Ý": 38,
43
- "Ă": 39,
44
- "Đ": 40,
45
- "Ĩ": 41,
46
- "Ũ": 42,
47
- "Ơ": 43,
48
- "Ư": 44,
49
- "Ạ": 45,
50
- "Ả": 46,
51
- "Ấ": 47,
52
- "Ầ": 48,
53
- "Ẩ": 49,
54
- "Ẫ": 50,
55
- "Ậ": 51,
56
- "Ắ": 52,
57
- "Ằ": 53,
58
- "Ẳ": 54,
59
- "Ẵ": 55,
60
- "Ặ": 56,
61
- "Ẹ": 57,
62
- "Ẻ": 58,
63
- "Ẽ": 59,
64
- "Ế": 60,
65
- "Ề": 61,
66
- "Ể": 62,
67
- "Ễ": 63,
68
- "Ệ": 64,
69
- "Ỉ": 65,
70
- "Ị": 66,
71
- "Ọ": 67,
72
- "Ỏ": 68,
73
- "Ố": 69,
74
- "Ồ": 70,
75
- "Ổ": 71,
76
- "Ỗ": 72,
77
- "Ộ": 73,
78
- "Ớ": 74,
79
- "Ờ": 75,
80
- "Ở": 76,
81
- "Ỡ": 77,
82
- "Ợ": 78,
83
- "Ụ": 79,
84
- "Ủ": 80,
85
- "Ứ": 81,
86
- "Ừ": 82,
87
- "Ử": 83,
88
- "Ữ": 84,
89
- "Ự": 85,
90
- "Ỳ": 86,
91
- "Ỵ": 87,
92
- "Ỷ": 88,
93
- "Ỹ": 89
94
  }
 
1
  {
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2
  "[PAD]": 91,
3
  "[UNK]": 90,
4
+ "a": 1,
5
+ "b": 2,
6
+ "c": 3,
7
+ "d": 4,
8
+ "e": 5,
9
+ "g": 6,
10
+ "h": 7,
11
+ "i": 8,
12
+ "k": 9,
13
+ "l": 10,
14
+ "m": 11,
15
+ "n": 12,
16
+ "o": 13,
17
+ "p": 14,
18
+ "q": 15,
19
+ "r": 16,
20
+ "s": 17,
21
+ "t": 18,
22
+ "u": 19,
23
+ "v": 20,
24
+ "x": 21,
25
+ "y": 22,
26
  "|": 0,
27
+ "à": 23,
28
+ "á": 24,
29
+ "â": 25,
30
+ "ã": 26,
31
+ "è": 27,
32
+ "é": 28,
33
+ "ê": 29,
34
+ "ì": 30,
35
+ "í": 31,
36
+ "ò": 32,
37
+ "ó": 33,
38
+ "ô": 34,
39
+ "õ": 35,
40
+ "ù": 36,
41
+ "ú": 37,
42
+ "ý": 38,
43
+ "ă": 39,
44
+ "đ": 40,
45
+ "ĩ": 41,
46
+ "ũ": 42,
47
+ "ơ": 43,
48
+ "ư": 44,
49
+ "ạ": 45,
50
+ "ả": 46,
51
+ "ấ": 47,
52
+ "ầ": 48,
53
+ "ẩ": 49,
54
+ "ẫ": 50,
55
+ "ậ": 51,
56
+ "ắ": 52,
57
+ "ằ": 53,
58
+ "ẳ": 54,
59
+ "ẵ": 55,
60
+ "ặ": 56,
61
+ "ẹ": 57,
62
+ "ẻ": 58,
63
+ "ẽ": 59,
64
+ "ế": 60,
65
+ "ề": 61,
66
+ "ể": 62,
67
+ "ễ": 63,
68
+ "ệ": 64,
69
+ "ỉ": 65,
70
+ "ị": 66,
71
+ "ọ": 67,
72
+ "ỏ": 68,
73
+ "ố": 69,
74
+ "ồ": 70,
75
+ "ổ": 71,
76
+ "ỗ": 72,
77
+ "ộ": 73,
78
+ "ớ": 74,
79
+ "ờ": 75,
80
+ "ở": 76,
81
+ "ỡ": 77,
82
+ "ợ": 78,
83
+ "ụ": 79,
84
+ "ủ": 80,
85
+ "ứ": 81,
86
+ "ừ": 82,
87
+ "ử": 83,
88
+ "ữ": 84,
89
+ "ự": 85,
90
+ "ỳ": 86,
91
+ "ỵ": 87,
92
+ "ỷ": 88,
93
+ "ỹ": 89
94
  }