mms-tts-prt / vocab.json
sanchit-gandhi's picture
Upload tokenizer
7039712
raw
history blame contribute delete
680 Bytes
{
" ": 38,
"'": 50,
"-": 48,
"0": 21,
"1": 0,
"2": 46,
"3": 25,
"4": 31,
"5": 26,
"6": 1,
"7": 47,
"8": 43,
"9": 30,
"_": 3,
"ก": 34,
"ค": 15,
"ง": 24,
"จ": 13,
"ซ": 53,
"ญ": 6,
"ด": 42,
"ต": 37,
"ท": 16,
"น": 12,
"บ": 18,
"ป": 9,
"พ": 10,
"ฟ": 14,
"ม": 54,
"ย": 29,
"ร": 5,
"ล": 51,
"ว": 20,
"ห": 36,
"อ": 7,
"ฮ": 35,
"ะ": 28,
"ั": 8,
"า": 32,
"ิ": 39,
"ี": 33,
"ึ": 44,
"ื": 11,
"ุ": 17,
"ู": 40,
"เ": 45,
"แ": 23,
"โ": 52,
"ใ": 22,
"ไ": 19,
"็": 41,
"้": 2,
"๊": 27,
"ํ": 49,
"‍": 4
}