mms-tts-cco / vocab.json
sanchit-gandhi's picture
Upload tokenizer
35dc579
raw
history blame
530 Bytes
{
" ": 46,
"'": 2,
"a": 1,
"b": 28,
"c": 16,
"d": 15,
"e": 4,
"f": 24,
"g": 11,
"h": 39,
"i": 3,
"j": 6,
"k": 45,
"l": 13,
"m": 20,
"n": 5,
"o": 10,
"p": 41,
"q": 27,
"r": 26,
"s": 14,
"t": 29,
"u": 12,
"v": 43,
"x": 44,
"y": 38,
"z": 42,
"|": 0,
"á": 22,
"é": 21,
"í": 23,
"ñ": 32,
"ó": 33,
"ø": 35,
"ú": 34,
"ü": 37,
"ǿ": 36,
"ɨ": 9,
"ʉ": 31,
"ˆ": 30,
"ˉ": 17,
"ˊ": 19,
"ˋ": 8,
"́": 25,
"̃": 18,
"̱": 7,
"—": 40
}