Thunpitcha
/

wav2vec2-base-thai-demo-google-colab

Automatic Speech Recognition

Generated from Trainer

Inference Endpoints

Model card Files Files and versions Metrics Training metrics Community

Thunpitcha commited on Jan 3

Commit

84d5f81

•

1 Parent(s): 606e0a6

Upload tokenizer

Files changed (1) hide show

vocab.json +65 -65

vocab.json CHANGED Viewed

@@ -1,70 +1,70 @@
 {
-  "'": 42,
   "[PAD]": 67,
   "[UNK]": 66,
-  "|": 65,
-  "ก": 31,
-  "ข": 36,
-  "ค": 28,
-  "ฆ": 23,
-  "ง": 22,
-  "จ": 61,
-  "ฉ": 7,
-  "ช": 39,
-  "ซ": 20,
-  "ญ": 11,
-  "ฎ": 41,
-  "ฏ": 50,
-  "ฐ": 52,
-  "ฑ": 26,
-  "ฒ": 44,
-  "ณ": 21,
-  "ด": 56,
-  "ต": 18,
-  "ถ": 40,
   "ท": 3,
-  "ธ": 29,
-  "น": 10,
-  "บ": 1,
-  "ป": 25,
-  "ผ": 17,
-  "ฝ": 15,
-  "พ": 12,
-  "ฟ": 55,
-  "ภ": 47,
-  "ม": 35,
-  "ย": 54,
-  "ร": 57,
-  "ฤ": 43,
-  "ล": 5,
-  "ว": 24,
-  "ศ": 38,
-  "ษ": 9,
-  "ส": 48,
-  "ห": 33,
-  "ฬ": 13,
-  "อ": 6,
-  "ฮ": 63,
-  "ะ": 2,
-  "ั": 59,
-  "า": 16,
-  "ำ": 62,
-  "ิ": 0,
-  "ี": 49,
-  "ึ": 46,
-  "ื": 30,
-  "ุ": 27,
-  "ู": 19,
-  "เ": 34,
-  "แ": 8,
-  "โ": 4,
-  "ใ": 45,
-  "ไ": 58,
-  "ๆ": 51,
-  "็": 14,
-  "่": 32,
-  "้": 53,
-  "๊": 37,
-  "๋": 60,
-  "์": 64
 }

 {
+  "'": 4,
   "[PAD]": 67,
   "[UNK]": 66,
+  "|": 40,
+  "ก": 0,
+  "ข": 53,
+  "ค": 51,
+  "ฆ": 17,
+  "ง": 26,
+  "จ": 52,
+  "ฉ": 25,
+  "ช": 33,
+  "ซ": 22,
+  "ญ": 16,
+  "ฎ": 23,
+  "ฏ": 65,
+  "ฐ": 43,
+  "ฑ": 21,
+  "ฒ": 56,
+  "ณ": 1,
+  "ด": 60,
+  "ต": 45,
+  "ถ": 62,
   "ท": 3,
+  "ธ": 63,
+  "น": 24,
+  "บ": 12,
+  "ป": 18,
+  "ผ": 61,
+  "ฝ": 11,
+  "พ": 31,
+  "ฟ": 29,
+  "ภ": 64,
+  "ม": 46,
+  "ย": 19,
+  "ร": 41,
+  "ฤ": 54,
+  "ล": 35,
+  "ว": 59,
+  "ศ": 44,
+  "ษ": 37,
+  "ส": 57,
+  "ห": 42,
+  "ฬ": 2,
+  "อ": 49,
+  "ฮ": 27,
+  "ะ": 47,
+  "ั": 15,
+  "า": 9,
+  "ำ": 48,
+  "ิ": 14,
+  "ี": 38,
+  "ึ": 20,
+  "ื": 28,
+  "ุ": 10,
+  "ู": 30,
+  "เ": 58,
+  "แ": 34,
+  "โ": 32,
+  "ใ": 50,
+  "ไ": 6,
+  "ๆ": 39,
+  "็": 13,
+  "่": 5,
+  "้": 55,
+  "๊": 7,
+  "๋": 8,
+  "์": 36
 }