colerobertson
/

wav2vec2-base-demo-colab

Automatic Speech Recognition

Generated from Trainer

Inference Endpoints

Model card Files Files and versions Metrics Training metrics Community

colerobertson commited on Mar 7

Commit

2dfb8d4

•

1 Parent(s): c05f3f1

Upload tokenizer

Files changed (3) hide show

added_tokens.json +2 -2
tokenizer_config.json +3 -3
vocab.json +48 -36

added_tokens.json CHANGED Viewed

@@ -1,4 +1,4 @@
 {
-  "</s>": 38,
-  "<s>": 37
 }

 {
+  "</s>": 50,
+  "<s>": 49
 }

tokenizer_config.json CHANGED Viewed

@@ -8,7 +8,7 @@
       "single_word": false,
       "special": false
     },
-    "36": {
       "content": "[UNK]",
       "lstrip": true,
       "normalized": false,
@@ -16,7 +16,7 @@
       "single_word": false,
       "special": false
     },
-    "37": {
       "content": "<s>",
       "lstrip": false,
       "normalized": false,
@@ -24,7 +24,7 @@
       "single_word": false,
       "special": true
     },
-    "38": {
       "content": "</s>",
       "lstrip": false,
       "normalized": false,

       "single_word": false,
       "special": false
     },
+    "48": {
       "content": "[UNK]",
       "lstrip": true,
       "normalized": false,
       "single_word": false,
       "special": false
     },
+    "49": {
       "content": "<s>",
       "lstrip": false,
       "normalized": false,
       "single_word": false,
       "special": true
     },
+    "50": {
       "content": "</s>",
       "lstrip": false,
       "normalized": false,

vocab.json CHANGED Viewed

@@ -1,39 +1,51 @@
 {
-  "'": 20,
-  "0": 17,
-  "1": 30,
-  "2": 7,
-  "3": 35,
-  "5": 24,
-  "9": 25,
   "[PAD]": 0,
-  "[UNK]": 36,
-  "a": 2,
-  "b": 21,
-  "c": 14,
-  "d": 3,
-  "e": 11,
-  "f": 10,
-  "g": 32,
-  "h": 4,
-  "i": 5,
-  "j": 28,
-  "k": 15,
-  "l": 27,
-  "m": 6,
-  "n": 23,
-  "o": 22,
-  "p": 1,
-  "q": 18,
-  "r": 16,
-  "s": 31,
-  "t": 34,
-  "u": 19,
-  "v": 9,
-  "w": 26,
-  "x": 33,
-  "y": 29,
-  "z": 13,
-  "|": 12,
-  "£": 8
 }

 {
   "[PAD]": 0,
+  "[UNK]": 48,
+  "a": 5,
+  "b": 16,
+  "d": 35,
+  "e": 36,
+  "f": 32,
+  "h": 24,
+  "i": 25,
+  "j": 14,
+  "k": 23,
+  "l": 12,
+  "m": 20,
+  "n": 46,
+  "o": 2,
+  "p": 8,
+  "r": 44,
+  "s": 15,
+  "t": 22,
+  "u": 17,
+  "w": 29,
+  "x": 7,
+  "y": 4,
+  "z": 37,
+  "|": 33,
+  "ç": 30,
+  "ŋ": 1,
+  "ɑ": 43,
+  "ɒ": 47,
+  "ɔ": 19,
+  "ə": 45,
+  "ɛ": 13,
+  "ɜ": 26,
+  "ɡ": 34,
+  "ɪ": 31,
+  "ɸ": 10,
+  "ɹ": 38,
+  "ɻ": 9,
+  "ʂ": 40,
+  "ʃ": 28,
+  "ʊ": 6,
+  "ʋ": 21,
+  "ʌ": 11,
+  "ʒ": 3,
+  "ʔ": 18,
+  "ʘ": 42,
+  "ʝ": 27,
+  "β": 39,
+  "θ": 41
 }