wav2vec2-darija / vocab.json
khaoulaoub's picture
Upload tokenizer
d499687
raw
history blame
418 Bytes
{
"[PAD]": 34,
"[UNK]": 33,
"|": 20,
"ء": 19,
"ا": 6,
"ب": 1,
"ة": 24,
"ت": 10,
"ث": 9,
"ج": 25,
"ح": 31,
"خ": 11,
"د": 29,
"ذ": 15,
"ر": 27,
"ز": 8,
"س": 21,
"ش": 5,
"ص": 3,
"ض": 14,
"ط": 16,
"ظ": 13,
"ع": 26,
"غ": 22,
"ـ": 12,
"ف": 23,
"ق": 32,
"ك": 7,
"ل": 18,
"م": 0,
"ن": 2,
"ه": 28,
"و": 30,
"ى": 4,
"ي": 17
}