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
a7a8578
•
1 Parent(s): bee669a

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -3
  2. tokenizer_config.json +4 -4
  3. vocab.json +83 -98
added_tokens.json CHANGED
@@ -1,5 +1,4 @@
1
  {
2
- "</s>": 100,
3
- "<s>": 99,
4
- "[PAD]": 98
5
  }
 
1
  {
2
+ "</s>": 85,
3
+ "<s>": 84
 
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "97": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "98": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "99": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": true,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "100": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": true,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "82": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "83": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "84": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": true,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "85": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": true,
vocab.json CHANGED
@@ -1,101 +1,86 @@
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
  }
 
1
  {
2
+ "'": 4,
3
+ "[PAD]": 83,
4
+ "[UNK]": 82,
5
+ "_": 57,
6
+ "c": 74,
7
+ "e": 47,
8
+ "i": 35,
9
+ "k": 63,
10
+ "l": 62,
11
+ "m": 30,
12
+ "n": 60,
13
+ "o": 56,
14
+ "r": 42,
15
+ "y": 15,
16
+ "|": 48,
17
+ "~": 16,
 
 
 
 
 
 
 
 
 
 
 
 
18
  "ก": 0,
19
+ "ข": 67,
20
+ "ค": 64,
21
+ "ฆ": 20,
22
+ "ง": 31,
23
+ "จ": 65,
24
+ "ฉ": 29,
25
+ "ช": 39,
26
+ "ซ": 26,
27
+ "ฌ": 24,
28
+ "ญ": 18,
29
+ "ฎ": 27,
30
+ "ฏ": 81,
31
+ "ฐ": 51,
32
+ "ฑ": 25,
33
+ "ฒ": 70,
34
+ "ณ": 1,
35
+ "ด": 76,
36
+ "ต": 53,
37
+ "ถ": 78,
38
+ "ท": 3,
39
+ "ธ": 79,
40
+ "น": 28,
41
+ "บ": 13,
42
+ "ป": 21,
43
+ "ผ": 77,
44
+ "ฝ": 12,
45
+ "พ": 38,
46
+ "ฟ": 34,
47
+ "ภ": 80,
48
+ "ม": 54,
49
+ "ย": 22,
50
+ "ร": 49,
51
+ "ฤ": 68,
52
+ "ล": 41,
53
+ "ว": 75,
54
+ "ศ": 52,
55
+ "ษ": 44,
56
+ "ส": 72,
57
+ "ห": 50,
58
+ "ฬ": 2,
59
+ "อ": 59,
60
+ "ฮ": 32,
61
+ "ะ": 55,
62
+ "ั": 19,
63
+ "า": 10,
64
+ "ำ": 58,
65
+ "ิ": 17,
66
+ "ี": 45,
67
+ "ึ": 23,
68
+ "ื": 33,
69
+ "ุ": 11,
70
+ "ู": 36,
71
+ "เ": 73,
72
+ "แ": 40,
73
+ "โ": 37,
74
+ "ใ": 61,
75
+ "ไ": 8,
76
+ "ๅ": 66,
77
+ "ๆ": 46,
78
+ "็": 14,
79
+ "่": 6,
80
+ "้": 69,
81
+ "๊": 7,
82
+ "๋": 9,
83
+ "์": 43,
84
+ "ํ": 5,
85
+ "’": 71
 
 
 
86
  }