Hugging Face's logo

Lasion
/
wav2vec2-common-voice-16_1_vi

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
wav2vec2
Generated from Trainer
Eval Results
Model card Files
xet
Metrics Community
Lasion commited on
Commit
f1d2e5d
·
verified ·
1 Parent(s): 7c1a3e1

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +93 -93
vocab.json CHANGED
@@ -1,98 +1,98 @@
1
  {
2
  "[PAD]": 95,
3
  "[UNK]": 94,
4
- "a": 64,
5
- "b": 38,
6
- "c": 28,
7
- "d": 29,
8
- "e": 91,
9
- "f": 92,
10
  "g": 90,
11
- "h": 20,
12
- "i": 56,
13
- "j": 15,
14
- "k": 43,
15
- "l": 76,
16
- "m": 74,
17
- "n": 55,
18
- "o": 36,
19
- "p": 68,
20
- "q": 31,
21
- "r": 67,
22
- "s": 89,
23
- "t": 75,
24
- "u": 54,
25
- "v": 73,
26
- "w": 66,
27
- "x": 49,
28
- "y": 4,
29
- "z": 58,
30
- "|": 46,
31
- "à": 2,
32
- "á": 13,
33
- "â": 48,
34
- "ã": 63,
35
- "è": 85,
36
- "é": 9,
37
- "ê": 57,
38
- "ì": 50,
39
- "í": 33,
40
- "ò": 30,
41
- "ó": 72,
42
- "ô": 82,
43
- "õ": 25,
44
- "ù": 17,
45
- "ú": 35,
46
- "ý": 51,
47
- "ă": 8,
48
- "đ": 37,
49
- "ĩ": 10,
50
- "ũ": 22,
51
- "ơ": 65,
52
- "ư": 3,
53
- "ạ": 84,
54
- "ả": 93,
55
- "ấ": 1,
56
- "ầ": 47,
57
- "ẩ": 61,
58
- "ẫ": 81,
59
- "ậ": 27,
60
- "ắ": 88,
61
- "ằ": 19,
62
- "ẳ": 79,
63
- "ẵ": 71,
64
- "ặ": 40,
65
- "ẹ": 32,
66
- "ẻ": 14,
67
- "ẽ": 45,
68
- "ế": 59,
69
- "ề": 41,
70
- "ể": 52,
71
- "ễ": 0,
72
- "ệ": 62,
73
- "ỉ": 83,
74
- "ị": 34,
75
- "ọ": 87,
76
- "ỏ": 18,
77
- "ố": 53,
78
- "ồ": 24,
79
- "ổ": 6,
80
- "ỗ": 69,
81
- "ộ": 70,
82
- "ớ": 80,
83
- "ờ": 5,
84
- "ở": 16,
85
- "ỡ": 11,
86
- "ợ": 86,
87
- "ụ": 12,
88
- "ủ": 7,
89
- "ứ": 26,
90
- "ừ": 39,
91
- "ử": 21,
92
- "ữ": 77,
93
- "ự": 23,
94
- "ỳ": 78,
95
- "ỵ": 44,
96
- "ỷ": 42,
97
- "ỹ": 60
98
  }
 
1
  {
2
  "[PAD]": 95,
3
  "[UNK]": 94,
4
+ "a": 86,
5
+ "b": 77,
6
+ "c": 58,
7
+ "d": 80,
8
+ "e": 37,
9
+ "f": 36,
10
  "g": 90,
11
+ "h": 8,
12
+ "i": 42,
13
+ "j": 71,
14
+ "k": 44,
15
+ "l": 27,
16
+ "m": 84,
17
+ "n": 62,
18
+ "o": 63,
19
+ "p": 34,
20
+ "q": 43,
21
+ "r": 38,
22
+ "s": 39,
23
+ "t": 14,
24
+ "u": 17,
25
+ "v": 50,
26
+ "w": 91,
27
+ "x": 61,
28
+ "y": 0,
29
+ "z": 15,
30
+ "|": 18,
31
+ "à": 25,
32
+ "á": 59,
33
+ "â": 70,
34
+ "ã": 55,
35
+ "è": 3,
36
+ "é": 82,
37
+ "ê": 56,
38
+ "ì": 29,
39
+ "í": 22,
40
+ "ò": 75,
41
+ "ó": 6,
42
+ "ô": 52,
43
+ "õ": 72,
44
+ "ù": 51,
45
+ "ú": 23,
46
+ "ý": 7,
47
+ "ă": 83,
48
+ "đ": 65,
49
+ "ĩ": 41,
50
+ "ũ": 21,
51
+ "ơ": 11,
52
+ "ư": 32,
53
+ "ạ": 92,
54
+ "ả": 26,
55
+ "ấ": 35,
56
+ "ầ": 76,
57
+ "ẩ": 9,
58
+ "ẫ": 20,
59
+ "ậ": 87,
60
+ "ắ": 2,
61
+ "ằ": 79,
62
+ "ẳ": 88,
63
+ "ẵ": 57,
64
+ "ặ": 45,
65
+ "ẹ": 49,
66
+ "ẻ": 47,
67
+ "ẽ": 53,
68
+ "ế": 48,
69
+ "ề": 89,
70
+ "ể": 54,
71
+ "ễ": 4,
72
+ "ệ": 5,
73
+ "ỉ": 31,
74
+ "ị": 66,
75
+ "ọ": 28,
76
+ "ỏ": 60,
77
+ "ố": 68,
78
+ "ồ": 16,
79
+ "ổ": 33,
80
+ "ỗ": 10,
81
+ "ộ": 73,
82
+ "ớ": 64,
83
+ "ờ": 19,
84
+ "ở": 85,
85
+ "ỡ": 46,
86
+ "ợ": 74,
87
+ "ụ": 93,
88
+ "ủ": 40,
89
+ "ứ": 13,
90
+ "ừ": 69,
91
+ "ử": 12,
92
+ "ữ": 81,
93
+ "ự": 78,
94
+ "ỳ": 67,
95
+ "ỵ": 1,
96
+ "ỷ": 24,
97
+ "ỹ": 30
98
  }