Hugging Face's logo

Sunbird
/
asr-mms-salt

Automatic Speech Recognition
Transformers
PyTorch
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Community
1
akera commited on
Commit
804ef6b
·
verified ·
1 Parent(s): 4fb9459

Update vocab.json

Browse files
Files changed (1) hide show
  1. vocab.json +290 -114
vocab.json CHANGED
@@ -1,128 +1,304 @@
1
  {
2
  "ach": {
3
- "=": 1,
4
- "[": 2,
5
- "[PAD]": 30,
6
- "[UNK]": 29,
7
- "a": 3,
8
- "b": 4,
9
- "c": 5,
10
- "d": 6,
11
- "e": 7,
12
- "f": 8,
13
- "g": 9,
14
- "h": 10,
15
- "i": 11,
16
- "j": 12,
17
- "k": 13,
18
- "l": 14,
19
- "m": 15,
20
- "n": 16,
21
- "o": 17,
22
- "p": 18,
23
- "q": 19,
24
- "r": 20,
25
- "s": 21,
26
- "t": 22,
27
- "u": 23,
28
- "v": 24,
29
- "w": 25,
30
- "x": 26,
31
- "y": 27,
32
- "z": 28,
33
- "|": 0
34
  },
35
- "lug": {
36
- "a": 1,
37
- "b": 2,
38
- "c": 3,
39
- "d": 4,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
40
  "e": 5,
41
- "f": 6,
42
- "g": 7,
43
- "h": 8,
44
  "i": 9,
45
- "j": 10,
46
- "k": 11,
47
- "l": 12,
48
- "m": 13,
49
- "n": 14,
50
- "o": 15,
51
- "p": 16,
52
- "q": 17,
53
- "r": 18,
54
- "s": 19,
55
- "t": 20,
56
- "u": 21,
57
- "v": 22,
58
  "w": 23,
59
- "x": 24,
60
- "y": 25,
61
- "z": 26,
62
- "|": 0,
63
- "[UNK]": 27,
64
- "[PAD]": 28
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
65
  },
66
- "teo": {
67
- "[PAD]": 28,
68
- "[UNK]": 27,
69
- "a": 1,
70
- "b": 2,
71
- "c": 3,
72
- "d": 4,
73
- "e": 5,
74
- "f": 6,
75
- "g": 7,
76
- "h": 8,
77
- "i": 9,
78
- "j": 10,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
79
  "k": 11,
80
- "l": 12,
81
  "m": 13,
82
- "n": 14,
83
- "o": 15,
84
- "p": 16,
85
- "q": 17,
86
- "r": 18,
87
  "s": 19,
88
- "t": 20,
89
- "u": 21,
90
- "v": 22,
91
- "w": 23,
92
- "x": 24,
93
- "y": 25,
94
- "z": 26,
95
- "|": 0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
96
  },
97
- "nyn": {
98
- "[PAD]": 28,
99
- "[UNK]": 27,
100
- "a": 1,
101
- "b": 2,
102
- "c": 3,
103
- "d": 4,
104
- "e": 5,
105
- "f": 6,
106
- "g": 7,
107
- "h": 8,
108
- "i": 9,
109
- "j": 10,
110
- "k": 11,
111
- "l": 12,
112
- "m": 13,
113
- "n": 14,
114
- "o": 15,
115
- "p": 16,
116
- "q": 17,
117
- "r": 18,
118
- "s": 19,
119
- "t": 20,
120
- "u": 21,
121
- "v": 22,
122
  "w": 23,
123
- "x": 24,
124
- "y": 25,
125
- "z": 26,
126
- "|": 0
 
127
  }
128
  }
 
1
  {
2
  "ach": {
3
+ "'": 27,
4
+ "-": 25,
5
+ "</s>": 2,
6
+ "<pad>": 0,
7
+ "<s>": 1,
8
+ "<unk>": 3,
9
+ "a": 6,
10
+ "b": 19,
11
+ "c": 17,
12
+ "d": 18,
13
+ "e": 8,
14
+ "f": 29,
15
+ "g": 20,
16
+ "h": 30,
17
+ "i": 7,
18
+ "j": 24,
19
+ "k": 9,
20
+ "l": 16,
21
+ "m": 11,
22
+ "n": 10,
23
+ "o": 5,
24
+ "p": 22,
25
+ "r": 21,
26
+ "s": 26,
27
+ "t": 13,
28
+ "u": 14,
29
+ "v": 28,
30
+ "w": 12,
31
+ "y": 15,
32
+ "|": 4,
33
+ "\u014b": 23
34
  },
35
+ "eng": {
36
+ "!": 75,
37
+ "\"": 62,
38
+ "$": 48,
39
+ "%": 47,
40
+ "&": 57,
41
+ "'": 31,
42
+ "+": 56,
43
+ ",": 44,
44
+ "-": 32,
45
+ ".": 37,
46
+ "/": 45,
47
+ "0": 27,
48
+ "1": 30,
49
+ "2": 35,
50
+ "3": 41,
51
+ "4": 39,
52
+ "5": 38,
53
+ "6": 40,
54
+ "7": 43,
55
+ "8": 42,
56
+ "9": 36,
57
+ ":": 46,
58
+ ";": 52,
59
+ "</s>": 2,
60
+ "<pad>": 0,
61
+ "<s>": 1,
62
+ "<unk>": 3,
63
+ "[": 59,
64
+ "]": 60,
65
+ "`": 92,
66
+ "a": 7,
67
+ "b": 24,
68
+ "c": 16,
69
+ "d": 15,
70
  "e": 5,
71
+ "f": 19,
72
+ "g": 21,
73
+ "h": 13,
74
  "i": 9,
75
+ "j": 29,
76
+ "k": 26,
77
+ "l": 14,
78
+ "m": 18,
79
+ "n": 10,
80
+ "o": 8,
81
+ "p": 20,
82
+ "q": 34,
83
+ "r": 12,
84
+ "s": 11,
85
+ "t": 6,
86
+ "u": 17,
87
+ "v": 25,
88
  "w": 23,
89
+ "x": 28,
90
+ "y": 22,
91
+ "z": 33,
92
+ "|": 4,
93
+ "\u00a3": 68,
94
+ "\u00a5": 53,
95
+ "\u00a7": 139,
96
+ "\u00b0": 58,
97
+ "\u00b2": 76,
98
+ "\u00df": 82,
99
+ "\u00e0": 84,
100
+ "\u00e1": 49,
101
+ "\u00e2": 81,
102
+ "\u00e3": 66,
103
+ "\u00e4": 78,
104
+ "\u00e5": 123,
105
+ "\u00e6": 110,
106
+ "\u00e7": 64,
107
+ "\u00e8": 79,
108
+ "\u00e9": 55,
109
+ "\u00ea": 88,
110
+ "\u00eb": 87,
111
+ "\u00ed": 50,
112
+ "\u00ee": 97,
113
+ "\u00ef": 85,
114
+ "\u00f0": 98,
115
+ "\u00f1": 83,
116
+ "\u00f2": 105,
117
+ "\u00f3": 67,
118
+ "\u00f4": 86,
119
+ "\u00f5": 69,
120
+ "\u00f6": 74,
121
+ "\u00f8": 91,
122
+ "\u00fa": 70,
123
+ "\u00fb": 99,
124
+ "\u00fc": 54,
125
+ "\u00fd": 112,
126
+ "\u00fe": 132,
127
+ "\u0101": 100,
128
+ "\u0103": 102,
129
+ "\u0107": 94,
130
+ "\u010d": 89,
131
+ "\u0113": 143,
132
+ "\u0119": 119,
133
+ "\u011b": 140,
134
+ "\u011f": 153,
135
+ "\u012b": 107,
136
+ "\u0131": 103,
137
+ "\u0142": 73,
138
+ "\u0144": 118,
139
+ "\u0148": 137,
140
+ "\u014d": 80,
141
+ "\u0151": 120,
142
+ "\u0153": 96,
143
+ "\u0159": 108,
144
+ "\u015f": 116,
145
+ "\u0161": 90,
146
+ "\u016b": 101,
147
+ "\u017e": 95,
148
+ "\u0219": 104,
149
+ "\u0259": 138,
150
+ "\u02bb": 115,
151
+ "\u0301": 93,
152
+ "\u0307": 77,
153
+ "\u03b1": 106,
154
+ "\u03ba": 109,
155
+ "\u03c0": 117,
156
+ "\u03c7": 141,
157
+ "\u0430": 144,
158
+ "\u0432": 124,
159
+ "\u0435": 125,
160
+ "\u0437": 130,
161
+ "\u0438": 113,
162
+ "\u0439": 127,
163
+ "\u043a": 114,
164
+ "\u043b": 126,
165
+ "\u043d": 128,
166
+ "\u044c": 131,
167
+ "\u044f": 129,
168
+ "\u05e0": 147,
169
+ "\u05e2": 148,
170
+ "\u1e43": 121,
171
+ "\u1ea1": 111,
172
+ "\u1ea3": 145,
173
+ "\u1ecb": 146,
174
+ "\u1ee5": 122,
175
+ "\u200b": 71,
176
+ "\u2013": 61,
177
+ "\u2014": 51,
178
+ "\u2018": 72,
179
+ "\u2019": 65,
180
+ "\u201d": 63,
181
+ "\u2261": 142,
182
+ "\u4eac": 149,
183
+ "\u5148": 135,
184
+ "\u5927": 151,
185
+ "\u5c1a": 134,
186
+ "\u65f6": 133,
187
+ "\u751f": 136,
188
+ "\u90fd": 150,
189
+ "\u962a": 152
190
  },
191
+ "lug": {
192
+ "!": 58,
193
+ "$": 52,
194
+ "%": 54,
195
+ "&": 66,
196
+ "'": 23,
197
+ "+": 55,
198
+ ",": 41,
199
+ "-": 44,
200
+ ".": 34,
201
+ "/": 48,
202
+ "0": 30,
203
+ "1": 31,
204
+ "2": 32,
205
+ "3": 40,
206
+ "4": 37,
207
+ "5": 36,
208
+ "6": 38,
209
+ "7": 43,
210
+ "8": 39,
211
+ "9": 35,
212
+ ":": 47,
213
+ ";": 53,
214
+ "</s>": 2,
215
+ "<pad>": 0,
216
+ "<s>": 1,
217
+ "<unk>": 3,
218
+ "?": 50,
219
+ "[": 70,
220
+ "]": 71,
221
+ "`": 73,
222
+ "a": 4,
223
+ "b": 12,
224
+ "c": 28,
225
+ "d": 22,
226
+ "e": 6,
227
+ "f": 25,
228
+ "g": 16,
229
+ "h": 29,
230
+ "i": 8,
231
+ "j": 24,
232
  "k": 11,
233
+ "l": 14,
234
  "m": 13,
235
+ "n": 10,
236
+ "o": 9,
237
+ "p": 26,
238
+ "q": 46,
239
+ "r": 20,
240
  "s": 19,
241
+ "t": 18,
242
+ "u": 7,
243
+ "v": 27,
244
+ "w": 17,
245
+ "x": 45,
246
+ "y": 15,
247
+ "z": 21,
248
+ "|": 5,
249
+ "\u00a3": 59,
250
+ "\u00a5": 64,
251
+ "\u00b0": 56,
252
+ "\u00b2": 69,
253
+ "\u00e1": 61,
254
+ "\u00e3": 65,
255
+ "\u00e7": 63,
256
+ "\u00ed": 62,
257
+ "\u00f5": 57,
258
+ "\u00fa": 68,
259
+ "\u00fc": 49,
260
+ "\u014b": 67,
261
+ "\u0300": 78,
262
+ "\u0301": 76,
263
+ "\u0313": 75,
264
+ "\u0314": 77,
265
+ "\u03b7": 74,
266
+ "\u0572": 72,
267
+ "\u2013": 51,
268
+ "\u2014": 60,
269
+ "\u2019": 33,
270
+ "\u201d": 42
271
  },
272
+ "teo": {
273
+ "'": 26,
274
+ "-": 27,
275
+ "</s>": 2,
276
+ "<pad>": 0,
277
+ "<s>": 1,
278
+ "<unk>": 3,
279
+ "a": 4,
280
+ "b": 21,
281
+ "c": 24,
282
+ "d": 17,
283
+ "e": 8,
284
+ "g": 25,
285
+ "i": 6,
286
+ "j": 22,
287
+ "k": 7,
288
+ "l": 15,
289
+ "m": 16,
290
+ "n": 10,
291
+ "o": 9,
292
+ "p": 19,
293
+ "r": 13,
294
+ "s": 14,
295
+ "t": 11,
296
+ "u": 12,
297
  "w": 23,
298
+ "y": 18,
299
+ "|": 5,
300
+ "\u014b": 20,
301
+ "\u2013": 29,
302
+ "\u2014": 28
303
  }
304
  }