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
8d69fe4
•
1 Parent(s): 203aaef

Upload tokenizer

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