Hugging Face's logo

Xenova
/
mms-tts-vie

Text-to-Speech
Transformers.js
ONNX
vits
text-to-audio
Model card Files Community
Xenova HF staff commited on
Commit
d99db3f
•
1 Parent(s): c622826

Upload tokenizer.json

Browse files
Files changed (1) hide show
  1. tokenizer.json +153 -0
tokenizer.json ADDED
@@ -0,0 +1,153 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": null,
4
+ "padding": null,
5
+ "added_tokens": [
6
+ {
7
+ "id": 95,
8
+ "content": "<unk>",
9
+ "single_word": false,
10
+ "lstrip": false,
11
+ "rstrip": false,
12
+ "normalized": false,
13
+ "special": true
14
+ }
15
+ ],
16
+ "normalizer": {
17
+ "type": "Sequence",
18
+ "normalizers": [
19
+ {
20
+ "type": "Lowercase"
21
+ },
22
+ {
23
+ "type": "Replace",
24
+ "pattern": {
25
+ "Regex": "[^\u1ee5xs\u00e8\u00ec\u1eb7\u1ee9\u01a1u\u01b0\u1eb1\u1ef3\u00e2cv\u1ec7\u1ed3_\u1ef5\u1ed9\u1ebbg\u1ea1\u0129\u1ea3\u00f5\u1ee1\u1eb3\u00f4aio\u1eeb\u1ef9\u1ed5\u00e0\u1edd\u00fd\u00f9\u2013\u1ec1\u1ec5\u1eb5\u1ecf'\u00f2\u1ed1q\u1ecd\u1ea9\u1ef1\u00e32\u1ecbe\u0111\u00f3\u1ea7\u00e9\u1edb\u1ebf\u1ee7\u1eedd\u1eef\u1eaf\u1ebd\u1eabm\u1ec3\u0169\u1ec9\u1eb9\u1ed7\u00edy\u00fa\u00e1pkt\u1ea5l\u1edf h\u1ef7\u1eadb\u0103n\u00ear\u1ee3\\-]"
26
+ },
27
+ "content": ""
28
+ },
29
+ {
30
+ "type": "Strip",
31
+ "strip_left": true,
32
+ "strip_right": true
33
+ },
34
+ {
35
+ "type": "Replace",
36
+ "pattern": {
37
+ "Regex": "(?=.)|(?<!^)$"
38
+ },
39
+ "content": "\u1ee5"
40
+ }
41
+ ]
42
+ },
43
+ "pre_tokenizer": {
44
+ "type": "Split",
45
+ "pattern": {
46
+ "Regex": ""
47
+ },
48
+ "behavior": "Isolated",
49
+ "invert": false
50
+ },
51
+ "post_processor": null,
52
+ "decoder": null,
53
+ "model": {
54
+ "vocab": {
55
+ "\u1ee5": 0,
56
+ "x": 1,
57
+ "s": 2,
58
+ "\u00e8": 3,
59
+ "\u00ec": 4,
60
+ "\u1eb7": 5,
61
+ "\u1ee9": 6,
62
+ "\u01a1": 7,
63
+ "u": 8,
64
+ "\u01b0": 9,
65
+ "\u1eb1": 10,
66
+ "\u1ef3": 11,
67
+ "\u00e2": 12,
68
+ "c": 13,
69
+ "v": 14,
70
+ "\u1ec7": 15,
71
+ "\u1ed3": 16,
72
+ "_": 17,
73
+ "\u1ef5": 18,
74
+ "\u1ed9": 19,
75
+ "\u1ebb": 20,
76
+ "g": 21,
77
+ "\u1ea1": 22,
78
+ "\u0129": 23,
79
+ "\u1ea3": 24,
80
+ "\u00f5": 25,
81
+ "\u1ee1": 26,
82
+ "\u1eb3": 27,
83
+ "\u00f4": 28,
84
+ "a": 29,
85
+ "i": 30,
86
+ "o": 31,
87
+ "\u1eeb": 32,
88
+ "\u1ef9": 33,
89
+ "\u1ed5": 34,
90
+ "\u00e0": 35,
91
+ "\u1edd": 36,
92
+ "\u00fd": 37,
93
+ "\u00f9": 38,
94
+ "\u2013": 39,
95
+ "\u1ec1": 40,
96
+ "\u1ec5": 41,
97
+ "\u1eb5": 42,
98
+ "\u1ecf": 43,
99
+ "'": 44,
100
+ "\u00f2": 45,
101
+ "\u1ed1": 46,
102
+ "q": 47,
103
+ "\u1ecd": 48,
104
+ "\u1ea9": 49,
105
+ "\u1ef1": 50,
106
+ "\u00e3": 51,
107
+ "2": 52,
108
+ "\u1ecb": 53,
109
+ "e": 54,
110
+ "\u0111": 55,
111
+ "\u00f3": 56,
112
+ "\u1ea7": 57,
113
+ "\u00e9": 58,
114
+ "\u1edb": 59,
115
+ "\u1ebf": 60,
116
+ "\u1ee7": 61,
117
+ "\u1eed": 62,
118
+ "d": 63,
119
+ "\u1eef": 64,
120
+ "\u1eaf": 65,
121
+ "\u1ebd": 66,
122
+ "\u1eab": 67,
123
+ "m": 68,
124
+ "\u1ec3": 69,
125
+ "\u0169": 70,
126
+ "\u1ec9": 71,
127
+ "\u1eb9": 72,
128
+ "\u1ed7": 73,
129
+ "\u00ed": 74,
130
+ "y": 75,
131
+ "\u00fa": 76,
132
+ "\u00e1": 77,
133
+ "p": 78,
134
+ "k": 79,
135
+ "t": 80,
136
+ "\u1ea5": 81,
137
+ "l": 82,
138
+ "\u1edf": 83,
139
+ " ": 84,
140
+ "h": 85,
141
+ "\u1ef7": 86,
142
+ "\u1ead": 87,
143
+ "b": 88,
144
+ "\u0103": 89,
145
+ "n": 90,
146
+ "\u00ea": 91,
147
+ "r": 92,
148
+ "\u1ee3": 93,
149
+ "-": 94,
150
+ "<unk>": 95
151
+ }
152
+ }
153
+ }