mms-tts-tye / vocab.json
sanchit-gandhi's picture
Upload tokenizer
1170057
raw
history blame
543 Bytes
{
" ": 18,
"'": 39,
"_": 2,
"a": 43,
"b": 21,
"c": 31,
"d": 11,
"e": 38,
"f": 3,
"g": 28,
"h": 14,
"i": 19,
"j": 23,
"k": 36,
"l": 1,
"m": 33,
"n": 35,
"o": 22,
"p": 46,
"r": 0,
"s": 4,
"t": 37,
"u": 30,
"v": 7,
"w": 29,
"y": 40,
"z": 26,
"à": 10,
"á": 9,
"ã": 12,
"è": 27,
"é": 32,
"ì": 8,
"í": 6,
"ò": 13,
"ó": 17,
"ù": 42,
"ú": 47,
"ĩ": 34,
"ń": 16,
"ũ": 24,
"ǹ": 20,
"ɔ": 41,
"ɛ": 44,
"̀": 25,
"́": 15,
"̃": 45,
"ḿ": 5
}