Hugging Face's logo

Thunpitcha
/
wav2vec2-base-thai-4-google-colab

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
Thunpitcha commited on
Commit
203aaef
•
1 Parent(s): b74139a

Upload tokenizer

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