Upload tokenizer

added_tokens.json

@@ -0,0 +1,4 @@
+{
+  "</s>": 724,
+  "<s>": 723
+}

special_tokens_map.json

@@ -0,0 +1,22 @@
+{
+  "additional_special_tokens": [
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    }
+  ],
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "pad_token": "[PAD]",
+  "unk_token": "[UNK]"
+}

tokenizer_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": true,
+  "do_lower_case": false,
+  "eos_token": "</s>",
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "[PAD]",
+  "replace_word_delimiter_char": " ",
+  "target_lang": "kor",
+  "tokenizer_class": "Wav2Vec2CTCTokenizer",
+  "unk_token": "[UNK]",
+  "word_delimiter_token": "|"
+}

vocab.json

@@ -0,0 +1,727 @@
+{
+  "kor": {
+    "[PAD]": 722,
+    "[UNK]": 721,
+    "k": 1,
+    "|": 0,
+    "’": 2,
+    "가": 3,
+    "각": 4,
+    "간": 5,
+    "갈": 6,
+    "감": 7,
+    "갑": 8,
+    "값": 9,
+    "갔": 10,
+    "강": 11,
+    "같": 12,
+    "갚": 13,
+    "갛": 14,
+    "개": 15,
+    "객": 16,
+    "거": 17,
+    "걱": 18,
+    "건": 19,
+    "걷": 20,
+    "걸": 21,
+    "검": 22,
+    "겁": 23,
+    "것": 24,
+    "게": 25,
+    "겠": 26,
+    "겨": 27,
+    "격": 28,
+    "견": 29,
+    "결": 30,
+    "겼": 31,
+    "경": 32,
+    "곁": 33,
+    "계": 34,
+    "고": 35,
+    "곡": 36,
+    "곤": 37,
+    "골": 38,
+    "곰": 39,
+    "곱": 40,
+    "곳": 41,
+    "공": 42,
+    "과": 43,
+    "관": 44,
+    "광": 45,
+    "괜": 46,
+    "괴": 47,
+    "굉": 48,
+    "교": 49,
+    "구": 50,
+    "국": 51,
+    "군": 52,
+    "굳": 53,
+    "굴": 54,
+    "권": 55,
+    "귀": 56,
+    "그": 57,
+    "근": 58,
+    "글": 59,
+    "금": 60,
+    "급": 61,
+    "긋": 62,
+    "기": 63,
+    "긴": 64,
+    "길": 65,
+    "김": 66,
+    "깊": 67,
+    "까": 68,
+    "깐": 69,
+    "깔": 70,
+    "깨": 71,
+    "꺼": 72,
+    "껄": 73,
+    "껑": 74,
+    "께": 75,
+    "꼈": 76,
+    "꼬": 77,
+    "꼭": 78,
+    "꼰": 79,
+    "꽉": 80,
+    "꾀": 81,
+    "꾸": 82,
+    "꾹": 83,
+    "꿈": 84,
+    "끄": 85,
+    "끈": 86,
+    "끊": 87,
+    "끌": 88,
+    "끓": 89,
+    "끔": 90,
+    "끝": 91,
+    "끼": 92,
+    "나": 93,
+    "낙": 94,
+    "난": 95,
+    "날": 96,
+    "남": 97,
+    "납": 98,
+    "났": 99,
+    "낮": 100,
+    "낯": 101,
+    "내": 102,
+    "냄": 103,
+    "냐": 104,
+    "냥": 105,
+    "너": 106,
+    "넉": 107,
+    "널": 108,
+    "넘": 109,
+    "네": 110,
+    "녀": 111,
+    "녁": 112,
+    "년": 113,
+    "녕": 114,
+    "노": 115,
+    "논": 116,
+    "놀": 117,
+    "놈": 118,
+    "농": 119,
+    "놓": 120,
+    "놔": 121,
+    "누": 122,
+    "눈": 123,
+    "눕": 124,
+    "느": 125,
+    "는": 126,
+    "늘": 127,
+    "늙": 128,
+    "능": 129,
+    "늦": 130,
+    "니": 131,
+    "닌": 132,
+    "님": 133,
+    "다": 134,
+    "닦": 135,
+    "단": 136,
+    "닫": 137,
+    "달": 138,
+    "닭": 139,
+    "닮": 140,
+    "담": 141,
+    "답": 142,
+    "닷": 143,
+    "당": 144,
+    "닿": 145,
+    "대": 146,
+    "댓": 147,
+    "더": 148,
+    "덕": 149,
+    "던": 150,
+    "덤": 151,
+    "덥": 152,
+    "덧": 153,
+    "덮": 154,
+    "데": 155,
+    "덴": 156,
+    "도": 157,
+    "독": 158,
+    "돈": 159,
+    "돌": 160,
+    "동": 161,
+    "됐": 162,
+    "되": 163,
+    "된": 164,
+    "될": 165,
+    "됨": 166,
+    "두": 167,
+    "둑": 168,
+    "둘": 169,
+    "둣": 170,
+    "둥": 171,
+    "뒤": 172,
+    "드": 173,
+    "득": 174,
+    "든": 175,
+    "듣": 176,
+    "들": 177,
+    "듬": 178,
+    "듯": 179,
+    "등": 180,
+    "디": 181,
+    "딘": 182,
+    "딪": 183,
+    "따": 184,
+    "딱": 185,
+    "딴": 186,
+    "땀": 187,
+    "땅": 188,
+    "때": 189,
+    "떠": 190,
+    "떡": 191,
+    "떤": 192,
+    "떨": 193,
+    "떻": 194,
+    "떼": 195,
+    "또": 196,
+    "똑": 197,
+    "똥": 198,
+    "뚜": 199,
+    "뚝": 200,
+    "뛰": 201,
+    "뜨": 202,
+    "뜩": 203,
+    "뜻": 204,
+    "띠": 205,
+    "띤": 206,
+    "라": 207,
+    "락": 208,
+    "란": 209,
+    "람": 210,
+    "랍": 211,
+    "랐": 212,
+    "랑": 213,
+    "랗": 214,
+    "래": 215,
+    "랫": 216,
+    "랬": 217,
+    "랴": 218,
+    "략": 219,
+    "러": 220,
+    "럭": 221,
+    "런": 222,
+    "럴": 223,
+    "럼": 224,
+    "렀": 225,
+    "렁": 226,
+    "렇": 227,
+    "레": 228,
+    "렌": 229,
+    "려": 230,
+    "력": 231,
+    "련": 232,
+    "렸": 233,
+    "령": 234,
+    "례": 235,
+    "로": 236,
+    "록": 237,
+    "론": 238,
+    "롭": 239,
+    "료": 240,
+    "룡": 241,
+    "루": 242,
+    "룩": 243,
+    "룽": 244,
+    "률": 245,
+    "르": 246,
+    "른": 247,
+    "를": 248,
+    "름": 249,
+    "릅": 250,
+    "릇": 251,
+    "릉": 252,
+    "리": 253,
+    "린": 254,
+    "릴": 255,
+    "림": 256,
+    "립": 257,
+    "릿": 258,
+    "마": 259,
+    "막": 260,
+    "만": 261,
+    "많": 262,
+    "말": 263,
+    "맘": 264,
+    "망": 265,
+    "맞": 266,
+    "맡": 267,
+    "매": 268,
+    "맹": 269,
+    "머": 270,
+    "먹": 271,
+    "먼": 272,
+    "멀": 273,
+    "멈": 274,
+    "멍": 275,
+    "메": 276,
+    "며": 277,
+    "면": 278,
+    "명": 279,
+    "몇": 280,
+    "모": 281,
+    "목": 282,
+    "몰": 283,
+    "몸": 284,
+    "몹": 285,
+    "못": 286,
+    "무": 287,
+    "묵": 288,
+    "문": 289,
+    "물": 290,
+    "뭇": 291,
+    "뭉": 292,
+    "뭐": 293,
+    "뭘": 294,
+    "므": 295,
+    "미": 296,
+    "민": 297,
+    "믿": 298,
+    "밀": 299,
+    "밑": 300,
+    "바": 301,
+    "박": 302,
+    "밖": 303,
+    "반": 304,
+    "받": 305,
+    "발": 306,
+    "밝": 307,
+    "밤": 308,
+    "밥": 309,
+    "방": 310,
+    "밭": 311,
+    "배": 312,
+    "백": 313,
+    "뱀": 314,
+    "버": 315,
+    "벅": 316,
+    "번": 317,
+    "벌": 318,
+    "법": 319,
+    "베": 320,
+    "벼": 321,
+    "벽": 322,
+    "변": 323,
+    "별": 324,
+    "볍": 325,
+    "병": 326,
+    "볕": 327,
+    "보": 328,
+    "복": 329,
+    "본": 330,
+    "볼": 331,
+    "봐": 332,
+    "봤": 333,
+    "부": 334,
+    "분": 335,
+    "불": 336,
+    "붉": 337,
+    "붐": 338,
+    "붙": 339,
+    "비": 340,
+    "빈": 341,
+    "빌": 342,
+    "빚": 343,
+    "빛": 344,
+    "빠": 345,
+    "빨": 346,
+    "빼": 347,
+    "뺄": 348,
+    "뺨": 349,
+    "뻘": 350,
+    "뼈": 351,
+    "뼉": 352,
+    "뽑": 353,
+    "뾰": 354,
+    "뿌": 355,
+    "뿐": 356,
+    "사": 357,
+    "산": 358,
+    "살": 359,
+    "삼": 360,
+    "삿": 361,
+    "상": 362,
+    "새": 363,
+    "색": 364,
+    "샘": 365,
+    "생": 366,
+    "서": 367,
+    "석": 368,
+    "섞": 369,
+    "선": 370,
+    "설": 371,
+    "섭": 372,
+    "섯": 373,
+    "섰": 374,
+    "성": 375,
+    "세": 376,
+    "센": 377,
+    "셔": 378,
+    "셨": 379,
+    "소": 380,
+    "속": 381,
+    "손": 382,
+    "솟": 383,
+    "송": 384,
+    "쇼": 385,
+    "수": 386,
+    "숙": 387,
+    "순": 388,
+    "술": 389,
+    "숨": 390,
+    "쉬": 391,
+    "쉴": 392,
+    "스": 393,
+    "슨": 394,
+    "슬": 395,
+    "슴": 396,
+    "습": 397,
+    "슷": 398,
+    "승": 399,
+    "시": 400,
+    "식": 401,
+    "신": 402,
+    "실": 403,
+    "싫": 404,
+    "심": 405,
+    "십": 406,
+    "싶": 407,
+    "싸": 408,
+    "싹": 409,
+    "쌀": 410,
+    "쌍": 411,
+    "쌓": 412,
+    "써": 413,
+    "썼": 414,
+    "쎄": 415,
+    "쏟": 416,
+    "쏴": 417,
+    "쑥": 418,
+    "쓰": 419,
+    "쓴": 420,
+    "쓸": 421,
+    "씀": 422,
+    "씨": 423,
+    "씩": 424,
+    "씬": 425,
+    "아": 426,
+    "악": 427,
+    "안": 428,
+    "앉": 429,
+    "않": 430,
+    "알": 431,
+    "암": 432,
+    "았": 433,
+    "앙": 434,
+    "앞": 435,
+    "애": 436,
+    "앤": 437,
+    "야": 438,
+    "약": 439,
+    "양": 440,
+    "얕": 441,
+    "어": 442,
+    "억": 443,
+    "언": 444,
+    "얹": 445,
+    "얻": 446,
+    "얼": 447,
+    "엄": 448,
+    "업": 449,
+    "없": 450,
+    "엇": 451,
+    "었": 452,
+    "엉": 453,
+    "엌": 454,
+    "엎": 455,
+    "에": 456,
+    "엔": 457,
+    "여": 458,
+    "역": 459,
+    "연": 460,
+    "열": 461,
+    "염": 462,
+    "였": 463,
+    "영": 464,
+    "옆": 465,
+    "예": 466,
+    "오": 467,
+    "옥": 468,
+    "온": 469,
+    "올": 470,
+    "옮": 471,
+    "옵": 472,
+    "옷": 473,
+    "와": 474,
+    "왔": 475,
+    "왜": 476,
+    "외": 477,
+    "요": 478,
+    "욕": 479,
+    "용": 480,
+    "우": 481,
+    "욱": 482,
+    "운": 483,
+    "울": 484,
+    "움": 485,
+    "웃": 486,
+    "워": 487,
+    "원": 488,
+    "월": 489,
+    "웠": 490,
+    "웬": 491,
+    "위": 492,
+    "유": 493,
+    "육": 494,
+    "윤": 495,
+    "으": 496,
+    "은": 497,
+    "을": 498,
+    "음": 499,
+    "읍": 500,
+    "의": 501,
+    "이": 502,
+    "인": 503,
+    "일": 504,
+    "읽": 505,
+    "잃": 506,
+    "임": 507,
+    "입": 508,
+    "잇": 509,
+    "있": 510,
+    "잊": 511,
+    "자": 512,
+    "작": 513,
+    "잔": 514,
+    "잘": 515,
+    "잠": 516,
+    "잡": 517,
+    "장": 518,
+    "재": 519,
+    "쟁": 520,
+    "저": 521,
+    "적": 522,
+    "전": 523,
+    "절": 524,
+    "젊": 525,
+    "점": 526,
+    "정": 527,
+    "제": 528,
+    "젯": 529,
+    "져": 530,
+    "졌": 531,
+    "조": 532,
+    "좀": 533,
+    "좁": 534,
+    "종": 535,
+    "좋": 536,
+    "죄": 537,
+    "죠": 538,
+    "주": 539,
+    "죽": 540,
+    "준": 541,
+    "줄": 542,
+    "줌": 543,
+    "중": 544,
+    "쥐": 545,
+    "즈": 546,
+    "즉": 547,
+    "즐": 548,
+    "증": 549,
+    "지": 550,
+    "직": 551,
+    "진": 552,
+    "질": 553,
+    "짐": 554,
+    "집": 555,
+    "짓": 556,
+    "징": 557,
+    "짖": 558,
+    "짜": 559,
+    "짝": 560,
+    "째": 561,
+    "쩍": 562,
+    "쪼": 563,
+    "쪽": 564,
+    "쫓": 565,
+    "쭉": 566,
+    "찌": 567,
+    "찔": 568,
+    "찢": 569,
+    "차": 570,
+    "착": 571,
+    "찬": 572,
+    "찮": 573,
+    "찰": 574,
+    "참": 575,
+    "창": 576,
+    "찾": 577,
+    "채": 578,
+    "책": 579,
+    "챘": 580,
+    "챙": 581,
+    "처": 582,
+    "척": 583,
+    "천": 584,
+    "철": 585,
+    "첨": 586,
+    "첫": 587,
+    "청": 588,
+    "체": 589,
+    "쳐": 590,
+    "쳤": 591,
+    "초": 592,
+    "총": 593,
+    "최": 594,
+    "추": 595,
+    "축": 596,
+    "춘": 597,
+    "출": 598,
+    "충": 599,
+    "취": 600,
+    "츠": 601,
+    "층": 602,
+    "치": 603,
+    "친": 604,
+    "칠": 605,
+    "침": 606,
+    "칭": 607,
+    "칵": 608,
+    "칼": 609,
+    "커": 610,
+    "컨": 611,
+    "컬": 612,
+    "켓": 613,
+    "켜": 614,
+    "켰": 615,
+    "코": 616,
+    "콘": 617,
+    "쾌": 618,
+    "쿵": 619,
+    "퀴": 620,
+    "크": 621,
+    "큰": 622,
+    "큼": 623,
+    "키": 624,
+    "킬": 625,
+    "타": 626,
+    "탄": 627,
+    "탓": 628,
+    "탕": 629,
+    "태": 630,
+    "택": 631,
+    "터": 632,
+    "턱": 633,
+    "털": 634,
+    "테": 635,
+    "텐": 636,
+    "토": 637,
+    "통": 638,
+    "퇴": 639,
+    "투": 640,
+    "툭": 641,
+    "튀": 642,
+    "트": 643,
+    "특": 644,
+    "튿": 645,
+    "틀": 646,
+    "틈": 647,
+    "티": 648,
+    "파": 649,
+    "판": 650,
+    "팔": 651,
+    "팠": 652,
+    "팡": 653,
+    "패": 654,
+    "퍼": 655,
+    "편": 656,
+    "폈": 657,
+    "평": 658,
+    "포": 659,
+    "폭": 660,
+    "표": 661,
+    "푹": 662,
+    "풀": 663,
+    "품": 664,
+    "프": 665,
+    "픈": 666,
+    "픔": 667,
+    "피": 668,
+    "필": 669,
+    "핏": 670,
+    "하": 671,
+    "학": 672,
+    "한": 673,
+    "할": 674,
+    "함": 675,
+    "합": 676,
+    "해": 677,
+    "핵": 678,
+    "햇": 679,
+    "했": 680,
+    "행": 681,
+    "향": 682,
+    "허": 683,
+    "헌": 684,
+    "헤": 685,
+    "혀": 686,
+    "혁": 687,
+    "현": 688,
+    "형": 689,
+    "혜": 690,
+    "호": 691,
+    "혹": 692,
+    "혼": 693,
+    "홍": 694,
+    "화": 695,
+    "확": 696,
+    "환": 697,
+    "활": 698,
+    "황": 699,
+    "회": 700,
+    "횡": 701,
+    "효": 702,
+    "후": 703,
+    "훈": 704,
+    "훌": 705,
+    "훔": 706,
+    "훤": 707,
+    "훨": 708,
+    "휘": 709,
+    "휙": 710,
+    "흉": 711,
+    "흔": 712,
+    "흘": 713,
+    "흙": 714,
+    "흥": 715,
+    "흩": 716,
+    "희": 717,
+    "히": 718,
+    "힌": 719,
+    "힘": 720
+  }
+}