Upload tokenizer

tokenizer.json

@@ -106,52 +106,53 @@
       "PAD": 1,
       "WORD_BOUNDARY": 2,
       "UTT_BOUNDARY": 3,
-      "k": 4,
-      "y": 5,
-      "m": 6,
-      "aɪ": 7,
-      "s": 8,
-      "t": 9,
-      "eɪ": 10,
-      "ŋ": 11,
-      "a": 12,
-      "i": 13,
-      "n": 14,
-      "ɛ": 15,
-      "æ": 16,
+      "m": 4,
+      "a": 5,
+      "s̪": 6,
+      "k": 7,
+      "ɛ": 8,
+      "ɾ": 9,
+      "u": 10,
+      "b": 11,
+      "e": 12,
+      "aʊ̯": 13,
+      "ɡ": 14,
+      "ɐ": 15,
+      "oɪ̯": 16,
       "z": 17,
-      "ɡ": 18,
-      "r": 19,
-      "v": 20,
-      "u": 21,
-      "ɾ": 22,
-      "d": 23,
-      "ʊ": 24,
-      "oɪ": 25,
-      "ɲ": 26,
-      "e": 27,
-      "f": 28,
-      "o": 29,
-      "p": 30,
-      "ʒ": 31,
-      "t̠ʃ": 32,
-      "d̠ʒ": 33,
-      "eʊ": 34,
-      "w": 35,
-      "aʊ": 36,
-      "j": 37,
-      "ə": 38,
-      "ũ": 39,
-      "ɐ̃": 40,
-      "l": 41,
-      "b": 42,
-      "x": 43,
-      "ɔ": 44,
-      "ʃ": 45,
-      "iʊ": 46,
-      "ɛʊ": 47,
-      "ɔɪ": 48,
-      "uɪ": 49
+      "i": 18,
+      "õ": 19,
+      "t̪": 20,
+      "eʊ̯": 21,
+      "n̪": 22,
+      "v": 23,
+      "d̪": 24,
+      "ɐ̃ʊ̯̃": 25,
+      "eɪ̯": 26,
+      "d̠ʒ": 27,
+      "ẽɪ̯̃": 28,
+      "p": 29,
+      "r": 30,
+      "ɔ": 31,
+      "o": 32,
+      "l": 33,
+      "ɐ̃": 34,
+      "ĩ": 35,
+      "f": 36,
+      "ɲ": 37,
+      "ũ": 38,
+      "uɪ̯": 39,
+      "w": 40,
+      "ʒ": 41,
+      "iʊ̯": 42,
+      "ʃ": 43,
+      "oʊ̯": 44,
+      "aɪ̯": 45,
+      "ɔɪ̯": 46,
+      "ɣ": 47,
+      "ɛɪ̯": 48,
+      "ɛʊ̯": 49,
+      "ɪ̯": 50
     },
     "unk_token": "UNK"
   }

vocab.json

@@ -1 +1 @@
-{"UNK":0,"PAD":1,"WORD_BOUNDARY":2,"UTT_BOUNDARY":3,"k":4,"y":5,"m":6,"aɪ":7,"s":8,"t":9,"eɪ":10,"ŋ":11,"a":12,"i":13,"n":14,"ɛ":15,"æ":16,"z":17,"ɡ":18,"r":19,"v":20,"u":21,"ɾ":22,"d":23,"ʊ":24,"oɪ":25,"ɲ":26,"e":27,"f":28,"o":29,"p":30,"ʒ":31,"t̠ʃ":32,"d̠ʒ":33,"eʊ":34,"w":35,"aʊ":36,"j":37,"ə":38,"ũ":39,"ɐ̃":40,"l":41,"b":42,"x":43,"ɔ":44,"ʃ":45,"iʊ":46,"ɛʊ":47,"ɔɪ":48,"uɪ":49}
+{"UNK":0,"PAD":1,"WORD_BOUNDARY":2,"UTT_BOUNDARY":3,"m":4,"a":5,"s̪":6,"k":7,"ɛ":8,"ɾ":9,"u":10,"b":11,"e":12,"aʊ̯":13,"ɡ":14,"ɐ":15,"oɪ̯":16,"z":17,"i":18,"õ":19,"t̪":20,"eʊ̯":21,"n̪":22,"v":23,"d̪":24,"ɐ̃ʊ̯̃":25,"eɪ̯":26,"d̠ʒ":27,"ẽɪ̯̃":28,"p":29,"r":30,"ɔ":31,"o":32,"l":33,"ɐ̃":34,"ĩ":35,"f":36,"ɲ":37,"ũ":38,"uɪ̯":39,"w":40,"ʒ":41,"iʊ̯":42,"ʃ":43,"oʊ̯":44,"aɪ̯":45,"ɔɪ̯":46,"ɣ":47,"ɛɪ̯":48,"ɛʊ̯":49,"ɪ̯":50}