parakeet-ctc-1.1b / vocab.json
nithinraok
add vocab file
bc5f3dc
{
"<unk>": 0,
"▁t": 1,
"▁th": 2,
"▁a": 3,
"▁i": 4,
"▁the": 5,
"re": 6,
"▁w": 7,
"▁s": 8,
"▁o": 9,
"in": 10,
"at": 11,
"er": 12,
"ou": 13,
"nd": 14,
"▁c": 15,
"▁b": 16,
"▁h": 17,
"on": 18,
"▁m": 19,
"▁f": 20,
"ing": 21,
"▁to": 22,
"en": 23,
"▁p": 24,
"▁and": 25,
"▁d": 26,
"es": 27,
"or": 28,
"an": 29,
"ll": 30,
"▁y": 31,
"▁l": 32,
"ed": 33,
"▁of": 34,
"▁in": 35,
"it": 36,
"is": 37,
"▁you": 38,
"▁that": 39,
"ar": 40,
"▁g": 41,
"▁n": 42,
"as": 43,
"om": 44,
"▁it": 45,
"ic": 46,
"ve": 47,
"▁e": 48,
"▁wh": 49,
"▁be": 50,
"us": 51,
"le": 52,
"al": 53,
"ion": 54,
"ow": 55,
"▁we": 56,
"▁re": 57,
"▁is": 58,
"ut": 59,
"ot": 60,
"ent": 61,
"▁on": 62,
"et": 63,
"▁ha": 64,
"ay": 65,
"ct": 66,
"▁he": 67,
"id": 68,
"▁for": 69,
"▁st": 70,
"ver": 71,
"ly": 72,
"ro": 73,
"ig": 74,
"▁so": 75,
"ld": 76,
"▁this": 77,
"ke": 78,
"▁u": 79,
"se": 80,
"all": 81,
"st": 82,
"ur": 83,
"ce": 84,
"ch": 85,
"im": 86,
"ith": 87,
"▁as": 88,
"▁k": 89,
"▁an": 90,
"▁was": 91,
"▁j": 92,
"▁with": 93,
"ir": 94,
"▁go": 95,
"ra": 96,
"▁do": 97,
"▁have": 98,
"▁li": 99,
"▁sh": 100,
"▁se": 101,
"▁they": 102,
"▁are": 103,
"am": 104,
"ht": 105,
"▁but": 106,
"ation": 107,
"▁not": 108,
"th": 109,
"▁r": 110,
"ally": 111,
"ad": 112,
"ust": 113,
"▁or": 114,
"▁com": 115,
"ould": 116,
"▁can": 117,
"ill": 118,
"▁ne": 119,
"ight": 120,
"▁ch": 121,
"▁de": 122,
"▁con": 123,
"▁at": 124,
"▁mo": 125,
"ant": 126,
"oo": 127,
"il": 128,
"▁me": 129,
"▁what": 130,
"▁there": 131,
"ter": 132,
"pe": 133,
"▁ab": 134,
"▁su": 135,
"ere": 136,
"ck": 137,
"▁pro": 138,
"▁al": 139,
"▁fr": 140,
"▁kn": 141,
"▁all": 142,
"ers": 143,
"▁like": 144,
"ge": 145,
"▁ex": 146,
"▁som": 147,
"ul": 148,
"▁your": 149,
"▁v": 150,
"pp": 151,
"use": 152,
"▁if": 153,
"ess": 154,
"ate": 155,
"est": 156,
"▁know": 157,
"out": 158,
"if": 159,
"▁just": 160,
"ment": 161,
"qu": 162,
"op": 163,
"ain": 164,
"▁one": 165,
"ol": 166,
"ri": 167,
"art": 168,
"very": 169,
"▁wor": 170,
"ive": 171,
"ist": 172,
"▁my": 173,
"nt": 174,
"ab": 175,
"▁from": 176,
"ort": 177,
"▁ma": 178,
"▁about": 179,
"res": 180,
"ity": 181,
"▁out": 182,
"▁bec": 183,
"▁le": 184,
"our": 185,
"od": 186,
"and": 187,
"ink": 188,
"ie": 189,
"▁up": 190,
"ind": 191,
"os": 192,
"un": 193,
"ause": 194,
"oug": 195,
"um": 196,
"▁some": 197,
"▁int": 198,
"▁by": 199,
"▁pl": 200,
"▁get": 201,
"el": 202,
"ard": 203,
"▁when": 204,
"▁don": 205,
"her": 206,
"▁will": 207,
"▁us": 208,
"▁would": 209,
"ook": 210,
"ies": 211,
"ich": 212,
"▁because": 213,
"▁think": 214,
"em": 215,
"▁pe": 216,
"▁his": 217,
"ack": 218,
"▁then": 219,
"▁our": 220,
"ide": 221,
"▁tim": 222,
"▁how": 223,
"ven": 224,
"▁tr": 225,
"▁who": 226,
"▁them": 227,
"ure": 228,
"▁ar": 229,
"▁ye": 230,
"▁more": 231,
"▁going": 232,
"ect": 233,
"▁sa": 234,
"▁cl": 235,
"▁had": 236,
"▁now": 237,
"▁which": 238,
"▁here": 239,
"ous": 240,
"▁their": 241,
"▁tw": 242,
"so": 243,
"▁has": 244,
"ud": 245,
"▁co": 246,
"▁ta": 247,
"ound": 248,
"▁were": 249,
"ast": 250,
"▁peop": 251,
"ough": 252,
"▁no": 253,
"▁really": 254,
"▁any": 255,
"▁people": 256,
"▁want": 257,
"▁she": 258,
"▁en": 259,
"▁fa": 260,
"▁te": 261,
"ame": 262,
"ine": 263,
"▁qu": 264,
"red": 265,
"▁im": 266,
"▁right": 267,
"ther": 268,
"▁act": 269,
"▁thing": 270,
"king": 271,
"ose": 272,
"▁ad": 273,
"▁see": 274,
"▁time": 275,
"▁these": 276,
"ci": 277,
"one": 278,
"▁say": 279,
"▁also": 280,
"▁fe": 281,
"per": 282,
"▁ag": 283,
"▁man": 284,
"ore": 285,
"▁un": 286,
"pt": 287,
"▁her": 288,
"▁look": 289,
"ong": 290,
"ice": 291,
"▁very": 292,
"ff": 293,
"ions": 294,
"▁comp": 295,
"▁did": 296,
"itt": 297,
"▁well": 298,
"▁other": 299,
"iv": 300,
"ase": 301,
"ree": 302,
"hing": 303,
"▁lo": 304,
"reat": 305,
"▁cont": 306,
"▁part": 307,
"▁into": 308,
"nder": 309,
"▁been": 310,
"are": 311,
"▁am": 312,
"ans": 313,
"▁sp": 314,
"▁two": 315,
"ue": 316,
"▁way": 317,
"age": 318,
"▁where": 319,
"ite": 320,
"▁dis": 321,
"▁than": 322,
"▁every": 323,
"▁pr": 324,
"▁po": 325,
"ag": 326,
"▁need": 327,
"ach": 328,
"iff": 329,
"ence": 330,
"pl": 331,
"own": 332,
"▁ac": 333,
"ble": 334,
"▁over": 335,
"iz": 336,
"▁work": 337,
"▁res": 338,
"▁make": 339,
"▁could": 340,
"▁off": 341,
"ually": 342,
"▁ro": 343,
"▁back": 344,
"able": 345,
"ip": 346,
"ry": 347,
"▁him": 348,
"▁cour": 349,
"ber": 350,
"▁pre": 351,
"▁fir": 352,
"▁spe": 353,
"ap": 354,
"ars": 355,
"▁diff": 356,
"ire": 357,
"▁somet": 358,
"▁imp": 359,
"▁those": 360,
"▁comm": 361,
"ance": 362,
"ick": 363,
"▁even": 364,
"ated": 365,
"way": 366,
"sel": 367,
"▁let": 368,
"▁br": 369,
"ty": 370,
"▁per": 371,
"int": 372,
"▁first": 373,
"▁thr": 374,
"▁under": 375,
"ah": 376,
"▁may": 377,
"▁cou": 378,
"▁new": 379,
"ress": 380,
"act": 381,
"▁gr": 382,
"ep": 383,
"▁said": 384,
"ations": 385,
"▁good": 386,
"ace": 387,
"ass": 388,
"▁does": 389,
"orm": 390,
"ish": 391,
"▁af": 392,
"ving": 393,
"co": 394,
"▁app": 395,
"▁lot": 396,
"▁things": 397,
"▁tra": 398,
"ittle": 399,
"▁bl": 400,
"▁little": 401,
"▁mu": 402,
"cess": 403,
"fe": 404,
"ome": 405,
"▁inc": 406,
"▁differe": 407,
"ary": 408,
"ical": 409,
"▁only": 410,
"ult": 411,
"▁again": 412,
"▁got": 413,
"ens": 414,
"▁gu": 415,
"▁kind": 416,
"▁much": 417,
"ord": 418,
"▁through": 419,
"ition": 420,
"ild": 421,
"▁down": 422,
"▁actually": 423,
"▁something": 424,
"ang": 425,
"ru": 426,
"ces": 427,
"▁fl": 428,
"ile": 429,
"ater": 430,
"▁ra": 431,
"▁take": 432,
"ict": 433,
"ign": 434,
"▁sc": 435,
"vel": 436,
"▁bet": 437,
"▁tal": 438,
"▁yeah": 439,
"▁use": 440,
"fore": 441,
"▁bu": 442,
"▁start": 443,
"ory": 444,
"be": 445,
"▁day": 446,
"wn": 447,
"xt": 448,
"ia": 449,
"ak": 450,
"▁after": 451,
"▁should": 452,
"▁fo": 453,
"▁ho": 454,
"▁hel": 455,
"▁ind": 456,
"▁uh": 457,
"na": 458,
"ial": 459,
"other": 460,
"▁ke": 461,
"▁call": 462,
"▁most": 463,
"▁ok": 464,
"▁different": 465,
"▁em": 466,
"ting": 467,
"ple": 468,
"▁being": 469,
"▁bo": 470,
"ning": 471,
"▁too": 472,
"ors": 473,
"▁happ": 474,
"ark": 475,
"og": 476,
"▁help": 477,
"▁rem": 478,
"du": 479,
"ction": 480,
"ood": 481,
"▁ser": 482,
"ether": 483,
"ious": 484,
"▁mean": 485,
"▁many": 486,
"▁court": 487,
"▁bel": 488,
"ade": 489,
"▁la": 490,
"ved": 491,
"▁des": 492,
"▁rec": 493,
"▁jo": 494,
"▁dec": 495,
"ves": 496,
"▁before": 497,
"▁put": 498,
"self": 499,
"▁point": 500,
"te": 501,
"▁ev": 502,
"form": 503,
"ents": 504,
"▁add": 505,
"ody": 506,
"thing": 507,
"▁case": 508,
"▁pers": 509,
"▁cons": 510,
"iss": 511,
"▁three": 512,
"oth": 513,
"▁ph": 514,
"▁come": 515,
"▁find": 516,
"▁why": 517,
"ull": 518,
"▁show": 519,
"▁bas": 520,
"▁great": 521,
"ily": 522,
"▁rel": 523,
"▁sm": 524,
"▁its": 525,
"▁fact": 526,
"▁pos": 527,
"ool": 528,
"ments": 529,
"ise": 530,
"nds": 531,
"ys": 532,
"▁try": 533,
"ual": 534,
"ful": 535,
"erm": 536,
"▁inter": 537,
"ons": 538,
"▁quest": 539,
"▁sub": 540,
"we": 541,
"vers": 542,
"▁supp": 543,
"▁feel": 544,
"▁same": 545,
"ub": 546,
"ates": 547,
"urn": 548,
"ert": 549,
"▁inv": 550,
"day": 551,
"▁rep": 552,
"igh": 553,
"▁sy": 554,
"▁inst": 555,
"▁long": 556,
"▁still": 557,
"▁okay": 558,
"ft": 559,
"ific": 560,
"atch": 561,
"ought": 562,
"ath": 563,
"▁own": 564,
"▁made": 565,
"ix": 566,
"ced": 567,
"ks": 568,
"lic": 569,
"▁wr": 570,
"de": 571,
"▁cr": 572,
"▁att": 573,
"▁ob": 574,
"▁world": 575,
"▁sure": 576,
"ward": 577,
"▁bit": 578,
"▁life": 579,
"▁person": 580,
"▁pres": 581,
"ph": 582,
"▁vide": 583,
"▁reg": 584,
"▁end": 585,
"ject": 586,
"ange": 587,
"▁fin": 588,
"ied": 589,
"pect": 590,
"▁didn": 591,
"▁around": 592,
"ian": 593,
"▁car": 594,
"ible": 595,
"▁sim": 596,
"ever": 597,
"▁sch": 598,
"ating": 599,
"▁pol": 600,
"▁set": 601,
"▁oh": 602,
"cy": 603,
"▁real": 604,
"▁import": 605,
"▁count": 606,
"▁um": 607,
"▁next": 608,
"cial": 609,
"les": 610,
"▁hu": 611,
"▁acc": 612,
"▁might": 613,
"▁ent": 614,
"▁doing": 615,
"▁ins": 616,
"▁gen": 617,
"▁play": 618,
"▁cle": 619,
"▁another": 620,
"ady": 621,
"ular": 622,
"ib": 623,
"ways": 624,
"ered": 625,
"ility": 626,
"ities": 627,
"▁op": 628,
"▁def": 629,
"▁years": 630,
"▁never": 631,
"ower": 632,
"ram": 633,
"▁tell": 634,
"▁sl": 635,
"onna": 636,
"ail": 637,
"ren": 638,
"ute": 639,
"▁gonna": 640,
"▁big": 641,
"▁give": 642,
"der": 643,
"ount": 644,
"▁ap": 645,
"kes": 646,
"▁state": 647,
"▁cor": 648,
"▁min": 649,
"ically": 650,
"▁mon": 651,
"▁fam": 652,
"▁important": 653,
"▁always": 654,
"▁high": 655,
"▁four": 656,
"▁gra": 657,
"▁ca": 658,
"▁stud": 659,
"▁dist": 660,
"▁talk": 661,
"▁num": 662,
"▁str": 663,
"▁today": 664,
"ract": 665,
"▁while": 666,
"ason": 667,
"▁iss": 668,
"▁sur": 669,
"▁char": 670,
"▁last": 671,
"oy": 672,
"ited": 673,
"▁exper": 674,
"▁place": 675,
"▁tri": 676,
"▁ear": 677,
"▁belie": 678,
"▁able": 679,
"▁underst": 680,
"▁che": 681,
"▁both": 682,
"ug": 683,
"▁doesn": 684,
"▁keep": 685,
"▁happen": 686,
"ings": 687,
"iew": 688,
"ather": 689,
"▁ass": 690,
"▁love": 691,
"ative": 692,
"av": 693,
"▁yes": 694,
"▁ele": 695,
"▁year": 696,
"▁such": 697,
"▁video": 698,
"ness": 699,
"▁el": 700,
"▁trans": 701,
"▁five": 702,
"▁produ": 703,
"ave": 704,
"erest": 705,
"als": 706,
"body": 707,
"cus": 708,
"▁found": 709,
"atter": 710,
"▁eff": 711,
"▁god": 712,
"▁used": 713,
"llow": 714,
"▁interest": 715,
"▁question": 716,
"hip": 717,
"▁bus": 718,
"▁ask": 719,
"▁exam": 720,
"▁prov": 721,
"lud": 722,
"▁form": 723,
"▁law": 724,
"ense": 725,
"▁child": 726,
"▁gl": 727,
"ne": 728,
"▁each": 729,
"▁understand": 730,
"▁care": 731,
"stem": 732,
"▁med": 733,
"▁maybe": 734,
"ably": 735,
"▁det": 736,
"▁coll": 737,
"its": 738,
"▁commun": 739,
"▁hand": 740,
"▁'": 741,
"▁ref": 742,
"▁lear": 743,
"▁done": 744,
"▁gener": 745,
"vern": 746,
"▁mr": 747,
"ween": 748,
"▁better": 749,
"▁between": 750,
"li": 751,
"blem": 752,
"▁system": 753,
"ertain": 754,
"▁school": 755,
"▁eas": 756,
"▁exp": 757,
"▁war": 758,
"ention": 759,
"▁ty": 760,
"▁govern": 761,
"ues": 762,
"▁problem": 763,
"▁plan": 764,
"ac": 765,
"▁conf": 766,
"▁course": 767,
"ouse": 768,
"▁mar": 769,
"▁stand": 770,
"▁sk": 771,
"▁seco": 772,
"uring": 773,
"▁ed": 774,
"▁mem": 775,
"ros": 776,
"cri": 777,
"▁thought": 778,
"cept": 779,
"▁partic": 780,
"▁test": 781,
"olog": 782,
"iness": 783,
"▁far": 784,
"led": 785,
"▁col": 786,
"▁looking": 787,
"▁read": 788,
"▁whether": 789,
"▁word": 790,
"me": 791,
"▁once": 792,
"ize": 793,
"▁home": 794,
"▁requ": 795,
"gg": 796,
"▁ide": 797,
"▁thank": 798,
"ures": 799,
"▁called": 800,
"▁cur": 801,
"▁water": 802,
"▁frie": 803,
"▁side": 804,
"▁best": 805,
"▁number": 806,
"oney": 807,
"▁turn": 808,
"ock": 809,
"▁eng": 810,
"▁top": 811,
"▁open": 812,
"ead": 813,
"▁everything": 814,
"▁term": 815,
"▁prob": 816,
"▁hard": 817,
"▁fun": 818,
"▁spec": 819,
"▁dire": 820,
"▁second": 821,
"▁pa": 822,
"▁build": 823,
"▁run": 824,
"▁sign": 825,
"▁reason": 826,
"▁inform": 827,
"▁watch": 828,
"ution": 829,
"▁few": 830,
"mo": 831,
"▁hum": 832,
"ision": 833,
"▁ext": 834,
"▁tog": 835,
"▁conc": 836,
"▁thous": 837,
"▁thousand": 838,
"▁support": 839,
"▁together": 840,
"▁six": 841,
"ps": 842,
"▁mark": 843,
"ics": 844,
"▁includ": 845,
"ef": 846,
"▁opp": 847,
"ident": 848,
"▁anything": 849,
"▁met": 850,
"▁bre": 851,
"▁jud": 852,
"▁away": 853,
"▁old": 854,
"▁prog": 855,
"ten": 856,
"▁book": 857,
"▁says": 858,
"▁seem": 859,
"▁contin": 860,
"▁process": 861,
"▁sing": 862,
"▁money": 863,
"▁having": 864,
"▁beg": 865,
"▁comple": 866,
"▁thir": 867,
"▁using": 868,
"▁ret": 869,
"ger": 870,
"▁head": 871,
"▁cre": 872,
"▁poss": 873,
"enty": 874,
"▁certain": 875,
"▁clear": 876,
"ines": 877,
"▁wee": 878,
"arch": 879,
"▁inf": 880,
"ont": 881,
"▁sit": 882,
"▁lead": 883,
"alth": 884,
"▁art": 885,
"ross": 886,
"▁pub": 887,
"▁without": 888,
"▁pret": 889,
"▁getting": 890,
"ient": 891,
"▁z": 892,
"▁wom": 893,
"▁power": 894,
"ational": 895,
"ner": 896,
"▁rest": 897,
"▁believe": 898,
"▁wa": 899,
"▁aut": 900,
"▁move": 901,
"aim": 902,
"▁sort": 903,
"idence": 904,
"▁creat": 905,
"▁expl": 906,
"▁name": 907,
"▁went": 908,
"▁eu": 909,
"▁change": 910,
"▁came": 911,
"▁pay": 912,
"ices": 913,
"▁sin": 914,
"▁pur": 915,
"▁pass": 916,
"▁whole": 917,
"▁house": 918,
"▁hund": 919,
"▁hundred": 920,
"▁pretty": 921,
"▁trying": 922,
"▁ple": 923,
"▁allow": 924,
"▁compan": 925,
"▁government": 926,
"▁small": 927,
"▁light": 928,
"▁bra": 929,
"▁stu": 930,
"aint": 931,
"▁ah": 932,
"▁prot": 933,
"ets": 934,
"▁cent": 935,
"velop": 936,
"▁family": 937,
"▁business": 938,
"ety": 939,
"▁making": 940,
"▁list": 941,
"▁experi": 942,
"eric": 943,
"▁follow": 944,
"ately": 945,
"▁probably": 946,
"▁appe": 947,
"▁serv": 948,
"▁val": 949,
"▁leg": 950,
"▁resp": 951,
"▁develop": 952,
"ready": 953,
"▁already": 954,
"▁sec": 955,
"ell": 956,
"▁saying": 957,
"ash": 958,
"▁hear": 959,
"▁loc": 960,
"▁adv": 961,
"▁pri": 962,
"ret": 963,
"▁lar": 964,
"▁beh": 965,
"▁must": 966,
"▁hon": 967,
"▁means": 968,
"ew": 969,
"▁par": 970,
"▁order": 971,
"▁mom": 972,
"gn": 973,
"▁though": 974,
"▁record": 975,
"▁miss": 976,
"▁dr": 977,
"▁es": 978,
"▁eight": 979,
"▁ever": 980,
"▁left": 981,
"▁example": 982,
"▁enough": 983,
"osed": 984,
"▁claim": 985,
"ank": 986,
"con": 987,
"▁americ": 988,
"▁information": 989,
"▁arg": 990,
"▁full": 991,
"nce": 992,
"▁consid": 993,
"▁working": 994,
"ature": 995,
"▁": 996,
"e": 997,
"t": 998,
"a": 999,
"o": 1000,
"i": 1001,
"n": 1002,
"s": 1003,
"r": 1004,
"h": 1005,
"l": 1006,
"d": 1007,
"u": 1008,
"c": 1009,
"m": 1010,
"y": 1011,
"w": 1012,
"g": 1013,
"f": 1014,
"p": 1015,
"b": 1016,
"v": 1017,
"k": 1018,
"'": 1019,
"j": 1020,
"x": 1021,
"q": 1022,
"z": 1023
}