Jarbas commited on
Commit
d3aaee5
·
verified ·
1 Parent(s): afdd5fa

fix tokenization: sorted vocab + phonemizer field

Browse files
Files changed (1) hide show
  1. coqui-fa-custom-glow-tts/config.json +134 -134
coqui-fa-custom-glow-tts/config.json CHANGED
@@ -23,141 +23,141 @@
23
  "<PAD>": 0,
24
  "<EOS>": 1,
25
  "<BOS>": 2,
26
- "ˈ": 3,
27
- "ˌ": 4,
28
- "ː": 5,
29
- "ˑ": 6,
30
- "p": 7,
31
- "b": 8,
32
- "t": 9,
33
- "d": 10,
34
- "ʈ": 11,
35
- "ɖ": 12,
36
- "c": 13,
37
- "ɟ": 14,
38
- "k": 15,
39
- "ɡ": 16,
40
- "q": 17,
41
- "ɢ": 18,
42
- "ʔ": 19,
43
- "ɴ": 20,
44
- "ŋ": 21,
45
- "ɲ": 22,
46
- "ɳ": 23,
47
- "n": 24,
48
- "ɱ": 25,
49
- "m": 26,
50
- "ʙ": 27,
51
- "r": 28,
52
- "ʀ": 29,
53
- "ⱱ": 30,
54
- "ɾ": 31,
55
- "ɽ": 32,
56
- "ɸ": 33,
57
- "β": 34,
58
- "f": 35,
59
- "v": 36,
60
- "θ": 37,
61
- "ð": 38,
62
- "s": 39,
63
- "z": 40,
64
- "ʃ": 41,
65
- "ʒ": 42,
66
- "ʂ": 43,
67
- "ʐ": 44,
68
- "ç": 45,
69
- "ʝ": 46,
70
- "x": 47,
71
- "ɣ": 48,
72
- "χ": 49,
73
- "ʁ": 50,
74
- "ħ": 51,
75
- "ʕ": 52,
76
- "h": 53,
77
- "ɦ": 54,
78
- "ɬ": 55,
79
- "ɮ": 56,
80
- "ʋ": 57,
81
- "ɹ": 58,
82
- "ɻ": 59,
83
- "j": 60,
84
- "ɰ": 61,
85
- "l": 62,
86
- "ɭ": 63,
87
- "ʎ": 64,
88
- "ʟ": 65,
89
- "a": 66,
90
- "e": 67,
91
- "g": 68,
92
- "i": 69,
93
- "o": 70,
94
- "u": 71,
95
  "w": 72,
96
- "y": 73,
97
- "ɪ": 74,
98
- "ʊ": 75,
99
- "̩": 76,
100
- "æ": 77,
101
- "ɑ": 78,
102
- "ɔ": 79,
103
- "ə": 80,
104
- "ɚ": 81,
105
- "ɛ": 82,
106
- "ɝ": 83,
107
- "ɨ": 84,
108
- "̃": 85,
109
- "ʉ": 86,
110
- "ʌ": 87,
111
- "ʍ": 88,
112
- "0": 89,
113
- "1": 90,
114
- "2": 91,
115
- "3": 92,
116
- "4": 93,
117
- "5": 94,
118
- "6": 95,
119
- "7": 96,
120
- "8": 97,
121
- "9": 98,
122
- "\"": 99,
123
- "#": 100,
124
- "$": 101,
125
- "%": 102,
126
- "*": 103,
127
- "+": 104,
128
- "/": 105,
129
- "=": 106,
130
- "A": 107,
131
- "B": 108,
132
- "C": 109,
133
- "D": 110,
134
- "E": 111,
135
- "F": 112,
136
- "G": 113,
137
- "H": 114,
138
- "I": 115,
139
- "J": 116,
140
- "K": 117,
141
- "L": 118,
142
- "M": 119,
143
- "N": 120,
144
- "O": 121,
145
- "P": 122,
146
- "R": 123,
147
- "S": 124,
148
- "T": 125,
149
- "U": 126,
150
- "V": 127,
151
- "W": 128,
152
- "X": 129,
153
- "Y": 130,
154
- "Z": 131,
155
- "[": 132,
156
- "]": 133,
157
- "^": 134,
158
- "_": 135,
159
- "{": 136,
160
- "}": 137,
161
  "!": 138,
162
  "(": 139,
163
  ")": 140,
 
23
  "<PAD>": 0,
24
  "<EOS>": 1,
25
  "<BOS>": 2,
26
+ "\"": 3,
27
+ "#": 4,
28
+ "$": 5,
29
+ "%": 6,
30
+ "*": 7,
31
+ "+": 8,
32
+ "/": 9,
33
+ "0": 10,
34
+ "1": 11,
35
+ "2": 12,
36
+ "3": 13,
37
+ "4": 14,
38
+ "5": 15,
39
+ "6": 16,
40
+ "7": 17,
41
+ "8": 18,
42
+ "9": 19,
43
+ "=": 20,
44
+ "A": 21,
45
+ "B": 22,
46
+ "C": 23,
47
+ "D": 24,
48
+ "E": 25,
49
+ "F": 26,
50
+ "G": 27,
51
+ "H": 28,
52
+ "I": 29,
53
+ "J": 30,
54
+ "K": 31,
55
+ "L": 32,
56
+ "M": 33,
57
+ "N": 34,
58
+ "O": 35,
59
+ "P": 36,
60
+ "R": 37,
61
+ "S": 38,
62
+ "T": 39,
63
+ "U": 40,
64
+ "V": 41,
65
+ "W": 42,
66
+ "X": 43,
67
+ "Y": 44,
68
+ "Z": 45,
69
+ "[": 46,
70
+ "]": 47,
71
+ "^": 48,
72
+ "_": 49,
73
+ "a": 50,
74
+ "b": 51,
75
+ "c": 52,
76
+ "d": 53,
77
+ "e": 54,
78
+ "f": 55,
79
+ "g": 56,
80
+ "h": 57,
81
+ "i": 58,
82
+ "j": 59,
83
+ "k": 60,
84
+ "l": 61,
85
+ "m": 62,
86
+ "n": 63,
87
+ "o": 64,
88
+ "p": 65,
89
+ "q": 66,
90
+ "r": 67,
91
+ "s": 68,
92
+ "t": 69,
93
+ "u": 70,
94
+ "v": 71,
95
  "w": 72,
96
+ "x": 73,
97
+ "y": 74,
98
+ "z": 75,
99
+ "{": 76,
100
+ "}": 77,
101
+ "æ": 78,
102
+ "ç": 79,
103
+ "ð": 80,
104
+ "ħ": 81,
105
+ "ŋ": 82,
106
+ "ɑ": 83,
107
+ "ɔ": 84,
108
+ "ɖ": 85,
109
+ "ə": 86,
110
+ "ɚ": 87,
111
+ "ɛ": 88,
112
+ "ɝ": 89,
113
+ "ɟ": 90,
114
+ "ɡ": 91,
115
+ "ɢ": 92,
116
+ "ɣ": 93,
117
+ "ɦ": 94,
118
+ "ɨ": 95,
119
+ "ɪ": 96,
120
+ "ɬ": 97,
121
+ "ɭ": 98,
122
+ "ɮ": 99,
123
+ "ɰ": 100,
124
+ "ɱ": 101,
125
+ "ɲ": 102,
126
+ "ɳ": 103,
127
+ "ɴ": 104,
128
+ "ɸ": 105,
129
+ "ɹ": 106,
130
+ "ɻ": 107,
131
+ "ɽ": 108,
132
+ "ɾ": 109,
133
+ "ʀ": 110,
134
+ "ʁ": 111,
135
+ "ʂ": 112,
136
+ "ʃ": 113,
137
+ "ʈ": 114,
138
+ "ʉ": 115,
139
+ "ʊ": 116,
140
+ "ʋ": 117,
141
+ "ʌ": 118,
142
+ "ʍ": 119,
143
+ "ʎ": 120,
144
+ "ʐ": 121,
145
+ "ʒ": 122,
146
+ "ʔ": 123,
147
+ "ʕ": 124,
148
+ "ʙ": 125,
149
+ "ʝ": 126,
150
+ "ʟ": 127,
151
+ "ˈ": 128,
152
+ "ˌ": 129,
153
+ "ː": 130,
154
+ "ˑ": 131,
155
+ "̃": 132,
156
+ "̩": 133,
157
+ "β": 134,
158
+ "θ": 135,
159
+ "χ": 136,
160
+ "ⱱ": 137,
161
  "!": 138,
162
  "(": 139,
163
  ")": 140,