mms-tts-cko / vocab.json
sanchit-gandhi's picture
Upload tokenizer
5dd4ccc
raw
history blame
398 Bytes
{
" ": 35,
"'": 34,
"a": 1,
"b": 7,
"c": 26,
"d": 16,
"e": 14,
"f": 18,
"g": 20,
"h": 25,
"i": 2,
"j": 22,
"k": 8,
"l": 23,
"m": 4,
"n": 3,
"o": 10,
"p": 24,
"r": 9,
"s": 12,
"t": 15,
"u": 6,
"v": 28,
"w": 17,
"y": 13,
"z": 27,
"|": 0,
"á": 21,
"é": 33,
"í": 30,
"ó": 29,
"ú": 31,
"ŋ": 19,
"ɔ": 11,
"ɛ": 5,
"́": 32
}