mms-tts-mil / vocab.json
sanchit-gandhi's picture
Upload tokenizer
4580bb8
raw
history blame
568 Bytes
{
" ": 5,
"_": 20,
"a": 0,
"b": 33,
"c": 7,
"d": 27,
"e": 17,
"f": 49,
"g": 1,
"h": 12,
"i": 26,
"j": 14,
"k": 37,
"l": 18,
"m": 29,
"n": 25,
"o": 10,
"p": 30,
"q": 6,
"r": 42,
"s": 9,
"t": 21,
"u": 46,
"v": 31,
"x": 48,
"y": 22,
"z": 15,
"à": 34,
"á": 3,
"è": 28,
"é": 41,
"ì": 40,
"í": 8,
"ñ": 38,
"ò": 45,
"ó": 24,
"ù": 35,
"ú": 2,
"ü": 16,
"ě": 47,
"ǎ": 36,
"ǐ": 11,
"ǒ": 43,
"ǔ": 13,
"ɨ": 23,
"̀": 4,
"́": 44,
"̌": 32,
"‐": 39,
"—": 19
}