Hugging Face's logo

TenzinGayche
/
wav2vec2-large-xlsr-uni

Automatic Speech Recognition
Transformers
PyTorch
Safetensors
wav2vec2
Inference Endpoints
Model card Files Community
1
TenzinGayche commited on
Commit
ba184d2
•
1 Parent(s): 03b7e42

Upload processor

Browse files
Files changed (2) hide show
  1. preprocessor_config.json +1 -0
  2. vocab.json +105 -105
preprocessor_config.json CHANGED
@@ -4,6 +4,7 @@
4
  "feature_size": 1,
5
  "padding_side": "right",
6
  "padding_value": 0.0,
 
7
  "return_attention_mask": true,
8
  "sampling_rate": 16000
9
  }
 
4
  "feature_size": 1,
5
  "padding_side": "right",
6
  "padding_value": 0.0,
7
+ "processor_class": "Wav2Vec2Processor",
8
  "return_attention_mask": true,
9
  "sampling_rate": 16000
10
  }
vocab.json CHANGED
@@ -2,109 +2,109 @@
2
  "[PAD]": 107,
3
  "[UNK]": 106,
4
  "a": 30,
5
- "b": 58,
6
- "c": 54,
7
- "d": 105,
8
- "e": 29,
9
- "f": 92,
10
- "g": 67,
11
- "h": 43,
12
- "i": 18,
13
- "j": 19,
14
- "k": 63,
15
- "l": 20,
16
- "m": 102,
17
- "n": 101,
18
- "o": 81,
19
- "p": 56,
20
- "r": 23,
21
- "s": 32,
22
- "t": 46,
23
- "u": 34,
24
- "v": 1,
25
- "w": 9,
26
- "y": 100,
27
- "z": 83,
28
- "|": 86,
29
- "་": 73,
30
- "༌": 8,
31
- "།": 74,
32
- "༑": 35,
33
- "༠": 6,
34
- "༡": 85,
35
- "༢": 60,
36
- "༣": 80,
37
- "༤": 75,
38
- "༥": 44,
39
- "༦": 59,
40
- "༧": 77,
41
- "༨": 33,
42
- "༩": 69,
43
- "ཀ": 22,
44
- "ཁ": 17,
45
- "ག": 7,
46
- "ང": 15,
47
- "ཅ": 10,
48
- "ཆ": 25,
49
- "ཇ": 103,
50
- "ཉ": 84,
51
- "ཊ": 76,
52
- "ཌ": 50,
53
- "ཎ": 16,
54
- "ཏ": 89,
55
- "ཐ": 37,
56
- "ད": 36,
57
- "ན": 72,
58
- "པ": 61,
59
- "ཕ": 3,
60
- "བ": 5,
61
- "མ": 11,
62
- "ཙ": 87,
63
- "ཚ": 91,
64
- "ཛ": 26,
65
- "ཝ": 66,
66
- "ཞ": 38,
67
- "ཟ": 49,
68
- "འ": 40,
69
- "ཡ": 64,
70
- "ར": 42,
71
- "ལ": 62,
72
- "ཤ": 55,
73
- "ས": 57,
74
- "ཧ": 39,
75
- "ཨ": 31,
76
- "ཱ": 99,
77
- "ི": 65,
78
- "ུ": 24,
79
- "ེ": 45,
80
- "ོ": 79,
81
- "ྐ": 27,
82
- "ྒ": 28,
83
- "ྔ": 53,
84
- "ྕ": 94,
85
- "ྗ": 82,
86
- "ྙ": 96,
87
- "ྜ": 47,
88
- "ྟ": 104,
89
- "ྡ": 97,
90
- "ྣ": 48,
91
- "ྤ": 0,
92
- "ྥ": 88,
93
- "ྦ": 98,
94
- "ྨ": 78,
95
- "ྩ": 93,
96
- "ྫ": 21,
97
- "ྭ": 41,
98
- "ྱ": 14,
99
- "ྲ": 4,
100
- "ླ": 51,
101
- "ྷ": 2,
102
- "“": 90,
103
- "”": 68,
104
- "五": 13,
105
- "傅": 12,
106
- "师": 70,
107
- "康": 95,
108
- "液": 52,
109
- "粮": 71
110
  }
 
2
  "[PAD]": 107,
3
  "[UNK]": 106,
4
  "a": 30,
5
+ "b": 55,
6
+ "c": 80,
7
+ "d": 85,
8
+ "e": 86,
9
+ "f": 71,
10
+ "g": 41,
11
+ "h": 37,
12
+ "i": 72,
13
+ "j": 77,
14
+ "k": 51,
15
+ "l": 16,
16
+ "m": 95,
17
+ "n": 4,
18
+ "o": 17,
19
+ "p": 53,
20
+ "r": 52,
21
+ "s": 33,
22
+ "t": 10,
23
+ "u": 87,
24
+ "v": 27,
25
+ "w": 46,
26
+ "y": 76,
27
+ "z": 13,
28
+ "|": 3,
29
+ "་": 78,
30
+ "༌": 90,
31
+ "།": 39,
32
+ "༑": 81,
33
+ "༠": 98,
34
+ "༡": 29,
35
+ "༢": 21,
36
+ "༣": 94,
37
+ "༤": 79,
38
+ "༥": 15,
39
+ "༦": 47,
40
+ "༧": 20,
41
+ "༨": 62,
42
+ "༩": 45,
43
+ "ཀ": 31,
44
+ "ཁ": 96,
45
+ "ག": 92,
46
+ "ང": 74,
47
+ "ཅ": 61,
48
+ "ཆ": 67,
49
+ "ཇ": 5,
50
+ "ཉ": 23,
51
+ "ཊ": 59,
52
+ "ཌ": 40,
53
+ "ཎ": 100,
54
+ "ཏ": 102,
55
+ "ཐ": 44,
56
+ "ད": 88,
57
+ "ན": 104,
58
+ "པ": 35,
59
+ "ཕ": 65,
60
+ "བ": 34,
61
+ "མ": 18,
62
+ "ཙ": 64,
63
+ "ཚ": 66,
64
+ "ཛ": 69,
65
+ "ཝ": 50,
66
+ "ཞ": 7,
67
+ "ཟ": 82,
68
+ "འ": 26,
69
+ "ཡ": 19,
70
+ "ར": 36,
71
+ "ལ": 68,
72
+ "ཤ": 22,
73
+ "ས": 97,
74
+ "ཧ": 56,
75
+ "ཨ": 14,
76
+ "ཱ": 63,
77
+ "ི": 42,
78
+ "ུ": 54,
79
+ "ེ": 70,
80
+ "ོ": 60,
81
+ "ྐ": 38,
82
+ "ྒ": 0,
83
+ "ྔ": 11,
84
+ "ྕ": 48,
85
+ "ྗ": 105,
86
+ "ྙ": 1,
87
+ "ྜ": 89,
88
+ "ྟ": 28,
89
+ "ྡ": 12,
90
+ "ྣ": 91,
91
+ "ྤ": 99,
92
+ "ྥ": 73,
93
+ "ྦ": 93,
94
+ "ྨ": 57,
95
+ "ྩ": 25,
96
+ "ྫ": 8,
97
+ "ྭ": 6,
98
+ "ྱ": 101,
99
+ "ྲ": 43,
100
+ "ླ": 58,
101
+ "ྷ": 83,
102
+ "“": 75,
103
+ "”": 2,
104
+ "五": 24,
105
+ "傅": 103,
106
+ "师": 32,
107
+ "康": 9,
108
+ "液": 84,
109
+ "粮": 49
110
  }