Upload tokenizer

tokenizer.json

@@ -113,68 +113,53 @@
       "PAD": 1,
       "WORD_BOUNDARY": 2,
       "UTT_BOUNDARY": 3,
-      "d": 4,
-      "uː": 5,
-      "j": 6,
-      "w": 7,
-      "ɔ": 8,
-      "n": 9,
-      "t": 10,
-      "ə": 11,
-      "l": 12,
-      "ʊ": 13,
-      "k": 14,
-      "æ": 15,
-      "ð": 16,
-      "ʌ": 17,
-      "ɪ": 18,
-      "s": 19,
-      "ɛ": 20,
-      "z": 21,
-      "iː": 22,
-      "ɹ": 23,
-      "f": 24,
-      "eɪ": 25,
-      "ɡ": 26,
-      "ɑ": 27,
-      "h": 28,
-      "p": 29,
-      "b": 30,
-      "i": 31,
-      "t̠ʃ": 32,
-      "aɪ": 33,
-      "θ": 34,
-      "ŋ": 35,
-      "m": 36,
-      "ɔɪ": 37,
-      "oʊ": 38,
-      "aʊ": 39,
-      "v": 40,
-      "ɜː": 41,
-      "d̠ʒ": 42,
-      "ʃ": 43,
-      "iə": 44,
+      "j": 4,
+      "ɛ": 5,
+      "h": 6,
+      "k": 7,
+      "ɑ": 8,
+      "m": 9,
+      "p": 10,
+      "aʊ": 11,
+      "n": 12,
+      "d": 13,
+      "z": 14,
+      "θ": 15,
+      "ɪ": 16,
+      "ŋ": 17,
+      "l": 18,
+      "aɪ": 19,
+      "s": 20,
+      "ɜː": 21,
+      "t": 22,
+      "w": 23,
+      "v": 24,
+      "ð": 25,
+      "æ": 26,
+      "ɔ": 27,
+      "ɹ": 28,
+      "ʌ": 29,
+      "f": 30,
+      "ə": 31,
+      "b": 32,
+      "iː": 33,
+      "eɪ": 34,
+      "oʊ": 35,
+      "d̠ʒ": 36,
+      "i": 37,
+      "uː": 38,
+      "iə": 39,
+      "ʊ": 40,
+      "ɡ": 41,
+      "t̠ʃ": 42,
+      "ɔɪ": 43,
+      "ʃ": 44,
       "ʒ": 45,
-      "ɑ̃": 46,
-      "r": 47,
-      "nʲ": 48,
-      "x": 49,
-      "ɬ": 50,
-      "ç": 51,
-      "e": 52,
-      "o": 53,
-      "ɛː": 54,
-      "ɪː": 55,
-      "u": 56,
-      "q": 57,
-      "tɕ": 58,
-      "tʰ": 59,
-      "ɯ": 60,
-      "r̩": 61,
-      "əʊ": 62,
-      "a": 63,
-      "ɒ": 64,
-      "eə": 65
+      "r": 46,
+      "x": 47,
+      "ɬ": 48,
+      "ɑ̃": 49,
+      "nʲ": 50
     },
     "unk_token": "UNK"
   }

vocab.json

@@ -1 +1 @@
-{"UNK":0,"PAD":1,"WORD_BOUNDARY":2,"UTT_BOUNDARY":3,"d":4,"uː":5,"j":6,"w":7,"ɔ":8,"n":9,"t":10,"ə":11,"l":12,"ʊ":13,"k":14,"æ":15,"ð":16,"ʌ":17,"ɪ":18,"s":19,"ɛ":20,"z":21,"iː":22,"ɹ":23,"f":24,"eɪ":25,"ɡ":26,"ɑ":27,"h":28,"p":29,"b":30,"i":31,"t̠ʃ":32,"aɪ":33,"θ":34,"ŋ":35,"m":36,"ɔɪ":37,"oʊ":38,"aʊ":39,"v":40,"ɜː":41,"d̠ʒ":42,"ʃ":43,"iə":44,"ʒ":45,"ɑ̃":46,"r":47,"nʲ":48,"x":49,"ɬ":50,"ç":51,"e":52,"o":53,"ɛː":54,"ɪː":55,"u":56,"q":57,"tɕ":58,"tʰ":59,"ɯ":60,"r̩":61,"əʊ":62,"a":63,"ɒ":64,"eə":65}
+{"UNK":0,"PAD":1,"WORD_BOUNDARY":2,"UTT_BOUNDARY":3,"j":4,"ɛ":5,"h":6,"k":7,"ɑ":8,"m":9,"p":10,"aʊ":11,"n":12,"d":13,"z":14,"θ":15,"ɪ":16,"ŋ":17,"l":18,"aɪ":19,"s":20,"ɜː":21,"t":22,"w":23,"v":24,"ð":25,"æ":26,"ɔ":27,"ɹ":28,"ʌ":29,"f":30,"ə":31,"b":32,"iː":33,"eɪ":34,"oʊ":35,"d̠ʒ":36,"i":37,"uː":38,"iə":39,"ʊ":40,"ɡ":41,"t̠ʃ":42,"ɔɪ":43,"ʃ":44,"ʒ":45,"r":46,"x":47,"ɬ":48,"ɑ̃":49,"nʲ":50}