Hugging Face's logo

Thunpitcha
/
wav2vec2-base-thai-5-google-colab

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
Thunpitcha commited on
Commit
5f4f8bd
•
1 Parent(s): b68f0af

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +84 -84
vocab.json CHANGED
@@ -1,91 +1,91 @@
1
  {
2
- "'": 38,
3
  "[PAD]": 88,
4
  "[UNK]": 87,
5
- "_": 53,
6
- "a": 73,
7
- "c": 72,
8
- "e": 83,
9
- "h": 21,
10
- "i": 6,
11
- "j": 18,
12
  "k": 36,
13
- "l": 78,
14
- "m": 70,
15
- "n": 47,
16
- "o": 22,
17
- "r": 82,
18
- "s": 7,
19
- "t": 79,
20
- "y": 65,
21
- "|": 71,
22
- "~": 24,
23
- "ก": 35,
24
- "ข": 54,
25
- "ค": 14,
26
- "ฆ": 10,
27
- "ง": 76,
28
- "จ": 28,
29
- "ฉ": 59,
30
- "ช": 81,
31
- "ซ": 33,
32
- "ฌ": 0,
33
- "ญ": 23,
34
- "ฎ": 30,
35
- "ฏ": 67,
36
- "ฐ": 12,
37
- "ฑ": 75,
38
- "ฒ": 42,
39
- "ณ": 77,
40
- "ด": 61,
41
- "ต": 2,
42
- "ถ": 57,
43
- "ท": 37,
44
- "ธ": 34,
45
- "น": 69,
46
- "บ": 29,
47
- "ป": 32,
48
- "ผ": 19,
49
- "ฝ": 55,
50
- "พ": 20,
51
  "ฟ": 56,
52
- "ภ": 26,
53
- "ม": 49,
54
- "ย": 8,
55
- "ร": 50,
56
- "ฤ": 39,
57
  "ล": 43,
58
- "ว": 27,
59
- "ศ": 9,
60
- "ษ": 5,
61
- "ส": 40,
62
- "ห": 85,
63
- "ฬ": 64,
64
- "อ": 1,
65
- "ฮ": 45,
66
- "ะ": 15,
67
- "ั": 25,
68
- "า": 17,
69
- "ำ": 51,
70
- "ิ": 4,
71
- "ี": 66,
72
- "ึ": 16,
73
- "ื": 62,
74
- "ุ": 3,
75
- "ู": 52,
76
- "เ": 63,
77
- "แ": 44,
78
- "โ": 74,
79
- "ใ": 41,
80
- "ไ": 31,
81
- "ๅ": 46,
82
- "ๆ": 13,
83
- "็": 84,
84
- "่": 60,
85
- "้": 86,
86
- "๊": 80,
87
- "๋": 68,
88
- "์": 11,
89
- "ํ": 48,
90
- "’": 58
91
  }
 
1
  {
2
+ "'": 2,
3
  "[PAD]": 88,
4
  "[UNK]": 87,
5
+ "_": 29,
6
+ "a": 45,
7
+ "c": 44,
8
+ "e": 52,
9
+ "h": 64,
10
+ "i": 34,
11
+ "j": 5,
12
  "k": 36,
13
+ "l": 8,
14
+ "m": 84,
15
+ "n": 23,
16
+ "o": 58,
17
+ "r": 50,
18
+ "s": 31,
19
+ "t": 63,
20
+ "y": 33,
21
+ "|": 22,
22
+ "~": 73,
23
+ "ก": 11,
24
+ "ข": 47,
25
+ "ค": 46,
26
+ "ฆ": 0,
27
+ "ง": 19,
28
+ "จ": 25,
29
+ "ฉ": 41,
30
+ "ช": 18,
31
+ "ซ": 77,
32
+ "ฌ": 6,
33
+ "ญ": 72,
34
+ "ฎ": 3,
35
+ "ฏ": 59,
36
+ "ฐ": 76,
37
+ "ฑ": 26,
38
+ "ฒ": 24,
39
+ "ณ": 54,
40
+ "ด": 42,
41
+ "ต": 14,
42
+ "ถ": 80,
43
+ "ท": 9,
44
+ "ธ": 55,
45
+ "น": 48,
46
+ "บ": 21,
47
+ "ป": 40,
48
+ "ผ": 60,
49
+ "ฝ": 85,
50
+ "พ": 53,
51
  "ฟ": 56,
52
+ "ภ": 67,
53
+ "ม": 62,
54
+ "ย": 10,
55
+ "ร": 37,
56
+ "ฤ": 61,
57
  "ล": 43,
58
+ "ว": 15,
59
+ "ศ": 79,
60
+ "ษ": 49,
61
+ "ส": 66,
62
+ "ห": 83,
63
+ "ฬ": 32,
64
+ "อ": 71,
65
+ "ฮ": 39,
66
+ "ะ": 35,
67
+ "ั": 86,
68
+ "า": 57,
69
+ "ำ": 12,
70
+ "ิ": 69,
71
+ "ี": 81,
72
+ "ึ": 68,
73
+ "ื": 70,
74
+ "ุ": 51,
75
+ "ู": 20,
76
+ "เ": 74,
77
+ "แ": 7,
78
+ "โ": 38,
79
+ "ใ": 28,
80
+ "ไ": 1,
81
+ "ๅ": 4,
82
+ "ๆ": 16,
83
+ "็": 27,
84
+ "่": 17,
85
+ "้": 65,
86
+ "๊": 82,
87
+ "๋": 78,
88
+ "์": 13,
89
+ "ํ": 75,
90
+ "’": 30
91
  }