mms-tts-box / vocab.json
sanchit-gandhi's picture
Upload tokenizer
4234b29
raw
history blame
556 Bytes
{
" ": 2,
"-": 23,
"6": 18,
"_": 13,
"a": 15,
"b": 3,
"c": 31,
"d": 0,
"e": 26,
"f": 34,
"h": 9,
"i": 47,
"k": 28,
"l": 21,
"m": 25,
"n": 30,
"o": 14,
"p": 35,
"r": 33,
"s": 6,
"t": 32,
"u": 40,
"v": 39,
"w": 37,
"y": 10,
"z": 43,
"à": 48,
"á": 44,
"ã": 24,
"è": 19,
"é": 17,
"ì": 38,
"í": 22,
"ò": 11,
"ó": 16,
"ù": 8,
"ú": 41,
"ĩ": 36,
"ń": 7,
"ũ": 46,
"ǹ": 20,
"ɓ": 29,
"ɔ": 12,
"ɛ": 45,
"ɲ": 27,
"̀": 42,
"́": 5,
"̃": 4,
"ṹ": 1
}