Hugging Face's logo

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

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
Thunpitcha commited on
Commit
bee669a
•
1 Parent(s): fc938cf

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +3 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +98 -67
added_tokens.json CHANGED
@@ -1,4 +1,5 @@
1
  {
2
- "</s>": 69,
3
- "<s>": 68
 
4
  }
 
1
  {
2
+ "</s>": 100,
3
+ "<s>": 99,
4
+ "[PAD]": 98
5
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "66": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "67": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "68": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": true,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "69": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": true,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "97": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "98": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "99": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": true,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "100": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": true,
vocab.json CHANGED
@@ -1,70 +1,101 @@
1
  {
2
- "'": 4,
3
- "[PAD]": 67,
4
- "[UNK]": 66,
5
- "|": 40,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
  "ก": 0,
7
- "ข": 53,
8
- "ค": 51,
9
- "ฆ": 17,
10
- "ง": 26,
11
- "จ": 52,
12
- "ฉ": 25,
13
- "ช": 33,
14
- "ซ": 22,
15
- "ญ": 16,
16
- "ฎ": 23,
17
- "ฏ": 65,
18
- "ฐ": 43,
19
- "ฑ": 21,
20
- "ฒ": 56,
21
- "ณ": 1,
22
- "ด": 60,
23
- "ต": 45,
24
- "ถ": 62,
25
- "ท": 3,
26
- "ธ": 63,
27
- "น": 24,
28
- "บ": 12,
29
- "ป": 18,
30
- "ผ": 61,
31
- "ฝ": 11,
32
- "พ": 31,
33
- "ฟ": 29,
34
- "ภ": 64,
35
- "ม": 46,
36
- "ย": 19,
37
- "ร": 41,
38
- "ฤ": 54,
39
- "ล": 35,
40
- "ว": 59,
41
- "ศ": 44,
42
- "ษ": 37,
43
- "ส": 57,
44
- "ห": 42,
45
- "ฬ": 2,
46
- "อ": 49,
47
- "ฮ": 27,
48
- "ะ": 47,
49
- "ั": 15,
50
- "า": 9,
51
- "ำ": 48,
52
- "ิ": 14,
53
- "ี": 38,
54
- "ึ": 20,
55
- "ื": 28,
56
- "ุ": 10,
57
- "ู": 30,
58
- "เ": 58,
59
- "แ": 34,
60
- "โ": 32,
61
- "ใ": 50,
62
- "ไ": 6,
63
- "ๆ": 39,
64
- "็": 13,
65
- "่": 5,
66
- "้": 55,
67
- "๊": 7,
68
- "๋": 8,
69
- "์": 36
 
 
 
 
 
 
 
70
  }
 
1
  {
2
+ "'": 3,
3
+ "[PAD]": 98,
4
+ "[UNK]": 97,
5
+ "_": 85,
6
+ "a": 32,
7
+ "b": 16,
8
+ "c": 43,
9
+ "e": 29,
10
+ "f": 71,
11
+ "g": 74,
12
+ "h": 38,
13
+ "i": 22,
14
+ "j": 70,
15
+ "k": 36,
16
+ "l": 87,
17
+ "m": 66,
18
+ "n": 34,
19
+ "o": 84,
20
+ "p": 50,
21
+ "r": 75,
22
+ "s": 37,
23
+ "t": 57,
24
+ "u": 44,
25
+ "w": 68,
26
+ "x": 1,
27
+ "y": 10,
28
+ "|": 78,
29
+ "~": 11,
30
  "ก": 0,
31
+ "ข": 92,
32
+ "ฃ": 7,
33
+ "ค": 88,
34
+ "ฆ": 59,
35
+ "ง": 19,
36
+ "จ": 89,
37
+ "ฉ": 65,
38
+ "ช": 25,
39
+ "ซ": 17,
40
+ "ฌ": 61,
41
+ "ญ": 13,
42
+ "ฎ": 63,
43
+ "ฏ": 48,
44
+ "ฐ": 81,
45
+ "ฑ": 62,
46
+ "ฒ": 40,
47
+ "ณ": 49,
48
+ "ด": 96,
49
+ "ต": 83,
50
+ "ถ": 45,
51
+ "ท": 2,
52
+ "ธ": 46,
53
+ "น": 64,
54
+ "บ": 8,
55
+ "ป": 60,
56
+ "ผ": 97,
57
+ "ฝ": 56,
58
+ "พ": 69,
59
+ "ฟ": 21,
60
+ "ภ": 47,
61
+ "ม": 30,
62
+ "ย": 14,
63
+ "ร": 79,
64
+ "ฤ": 93,
65
+ "ล": 73,
66
+ "ว": 95,
67
+ "ศ": 82,
68
+ "ษ": 27,
69
+ "ส": 94,
70
+ "ห": 80,
71
+ "ฬ": 51,
72
+ "อ": 86,
73
+ "ฮ": 67,
74
+ "ฯ": 91,
75
+ "ะ": 31,
76
+ "ั": 58,
77
+ "า": 55,
78
+ "ำ": 33,
79
+ "ิ": 12,
80
+ "ี": 77,
81
+ "ึ": 15,
82
+ "ื": 20,
83
+ "ุ": 6,
84
+ "ู": 23,
85
+ "เ": 42,
86
+ "แ": 72,
87
+ "โ": 24,
88
+ "ใ": 35,
89
+ "ไ": 54,
90
+ "ๅ": 90,
91
+ "ๆ": 28,
92
+ "็": 9,
93
+ "่": 4,
94
+ "้": 39,
95
+ "๊": 53,
96
+ "๋": 5,
97
+ "์": 76,
98
+ "ํ": 52,
99
+ "—": 26,
100
+ "’": 41
101
  }