wav2vec2-base / vocab.json
HuyenNguyen's picture
Upload tokenizer
9ba62ac
raw
history blame
1.18 kB
{
"'": 45,
"[PAD]": 96,
"[UNK]": 95,
"a": 51,
"b": 3,
"c": 18,
"d": 25,
"e": 54,
"f": 11,
"g": 94,
"h": 5,
"i": 74,
"j": 37,
"k": 7,
"l": 31,
"m": 16,
"n": 21,
"o": 89,
"p": 63,
"q": 34,
"r": 70,
"s": 38,
"t": 82,
"u": 19,
"v": 40,
"w": 39,
"x": 88,
"y": 57,
"z": 26,
"|": 29,
"à": 66,
"á": 44,
"â": 83,
"ã": 61,
"è": 77,
"é": 93,
"ê": 30,
"ì": 56,
"í": 73,
"ò": 8,
"ó": 46,
"ô": 71,
"õ": 6,
"ù": 67,
"ú": 80,
"ý": 50,
"ă": 9,
"đ": 91,
"ĩ": 49,
"ũ": 76,
"ơ": 84,
"ư": 24,
"ạ": 27,
"ả": 22,
"ấ": 36,
"ầ": 12,
"ẩ": 59,
"ẫ": 2,
"ậ": 79,
"ắ": 42,
"ằ": 28,
"ẳ": 55,
"ẵ": 33,
"ặ": 52,
"ẹ": 35,
"ẻ": 13,
"ẽ": 41,
"ế": 65,
"ề": 4,
"ể": 15,
"ễ": 86,
"ệ": 58,
"ỉ": 90,
"ị": 64,
"ọ": 10,
"ỏ": 68,
"ố": 23,
"ồ": 47,
"ổ": 81,
"ỗ": 1,
"ộ": 14,
"ớ": 60,
"ờ": 17,
"ở": 20,
"ỡ": 75,
"ợ": 43,
"ụ": 69,
"ủ": 78,
"ứ": 62,
"ừ": 0,
"ử": 72,
"ữ": 87,
"ự": 53,
"ỳ": 32,
"ỵ": 92,
"ỷ": 85,
"ỹ": 48
}