mms-tts-ncu / vocab.json
sanchit-gandhi's picture
Upload tokenizer
33a4337
raw
history blame
341 Bytes
{
" ": 30,
"'": 26,
"-": 17,
"a": 3,
"b": 10,
"d": 24,
"e": 2,
"f": 20,
"g": 22,
"h": 27,
"i": 13,
"k": 9,
"l": 23,
"m": 7,
"n": 5,
"o": 4,
"p": 21,
"r": 12,
"s": 11,
"t": 19,
"u": 18,
"w": 16,
"y": 8,
"|": 0,
"ò": 25,
"ŋ": 15,
"ɔ": 6,
"ɛ": 14,
"ʻ": 28,
"ʼ": 29,
"̱": 1
}