mms-tts-bom / vocab.json
sanchit-gandhi's picture
Upload tokenizer
f439cac
raw
history blame contribute delete
698 Bytes
{
" ": 20,
"'": 36,
"-": 18,
"_": 35,
"a": 4,
"b": 56,
"c": 39,
"d": 22,
"e": 47,
"f": 48,
"g": 3,
"h": 2,
"i": 17,
"j": 37,
"k": 6,
"l": 25,
"m": 46,
"n": 31,
"o": 44,
"p": 26,
"r": 29,
"s": 24,
"t": 51,
"u": 28,
"v": 30,
"w": 21,
"y": 11,
"z": 23,
"à": 16,
"á": 34,
"â": 33,
"è": 27,
"é": 12,
"ê": 15,
"ì": 52,
"í": 14,
"î": 50,
"ò": 10,
"ó": 5,
"ô": 59,
"ù": 38,
"ú": 9,
"û": 45,
"ā": 57,
"ē": 60,
"ě": 55,
"ī": 49,
"ō": 0,
"ū": 40,
"ǎ": 53,
"ǐ": 58,
"ǒ": 1,
"ǔ": 8,
"ɔ": 32,
"ɛ": 7,
"̀": 54,
"́": 13,
"̂": 42,
"̄": 43,
"̌": 41,
"—": 19
}