mms-tts-cuc / vocab.json
sanchit-gandhi's picture
Upload tokenizer
07849af
raw
history blame
441 Bytes
{
" ": 39,
"1": 15,
"2": 7,
"3": 1,
"4": 4,
"5": 21,
"a": 3,
"b": 29,
"c": 13,
"d": 25,
"e": 8,
"f": 32,
"g": 20,
"h": 6,
"i": 2,
"j": 9,
"k": 38,
"l": 12,
"m": 17,
"n": 5,
"o": 10,
"p": 26,
"q": 19,
"r": 23,
"s": 16,
"t": 14,
"u": 11,
"v": 35,
"x": 37,
"y": 24,
"z": 36,
"|": 0,
"á": 31,
"é": 33,
"í": 30,
"ñ": 27,
"ó": 34,
"ú": 22,
"ŋ": 28,
"ˉ": 18
}