Hugging Face's logo

mattbonnell
/
wav2vec2-base-wonders-phonemes

Automatic Speech Recognition
Transformers
Safetensors
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Community
mattbonnell commited on
Commit
e3db447
·
verified ·
1 Parent(s): ec0afae

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +42 -45
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 48,
3
- "<s>": 47
4
  }
 
1
  {
2
+ "</s>": 45,
3
+ "<s>": 44
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "45": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "46": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "47": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "48": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "42": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "43": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "44": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "45": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -1,49 +1,46 @@
1
  {
2
- "[PAD]": 46,
3
- "[UNK]": 45,
4
- "a": 41,
5
- "b": 16,
6
  "d": 28,
7
- "e": 43,
8
- "f": 40,
9
- "h": 44,
10
- "i": 0,
11
- "j": 7,
12
- "k": 22,
13
- "l": 11,
14
- "m": 25,
15
- "n": 5,
16
- "o": 3,
17
- "p": 20,
18
- "s": 19,
19
- "t": 24,
20
- "u": 42,
21
- "v": 13,
22
- "w": 6,
23
- "z": 37,
24
- "|": 2,
25
- "æ": 31,
26
- "ð": 30,
27
- "ŋ": 36,
28
- "ɐ": 26,
29
- "ɑ": 21,
30
- "ɔ": 14,
31
- "ə": 15,
32
- "ɚ": 34,
33
- "ɛ": 18,
34
- "ɜ": 27,
35
  "ɡ": 32,
36
- "ɣ": 17,
37
- "ɪ": 39,
38
- "ɹ": 38,
39
- "ɾ": 10,
40
- "ʃ": 9,
41
- "ʊ": 29,
42
- "ʌ": 33,
43
- "ʒ": 23,
44
- "ʔ": 12,
45
- "ː": 1,
46
- "̩": 4,
47
- "θ": 35,
48
- "ᵻ": 8
49
  }
 
1
  {
2
+ "[PAD]": 43,
3
+ "[UNK]": 42,
4
+ "a": 6,
5
+ "b": 20,
6
  "d": 28,
7
+ "e": 14,
8
+ "f": 19,
9
+ "h": 27,
10
+ "i": 3,
11
+ "j": 10,
12
+ "k": 23,
13
+ "l": 30,
14
+ "m": 18,
15
+ "n": 9,
16
+ "o": 38,
17
+ "p": 13,
18
+ "s": 31,
19
+ "t": 11,
20
+ "u": 24,
21
+ "v": 12,
22
+ "w": 39,
23
+ "z": 7,
24
+ "|": 17,
25
+ "æ": 33,
26
+ "ð": 0,
27
+ "ŋ": 15,
28
+ "ɐ": 36,
29
+ "ɑ": 35,
30
+ "ɔ": 29,
31
+ "ə": 2,
32
+ "ɚ": 21,
33
+ "ɛ": 5,
34
+ "ɜ": 41,
35
  "ɡ": 32,
36
+ "ɪ": 37,
37
+ "ɹ": 34,
38
+ "ɾ": 25,
39
+ "ʃ": 1,
40
+ "ʊ": 22,
41
+ "ʌ": 40,
42
+ "ʒ": 8,
43
+ "ː": 16,
44
+ "θ": 26,
45
+ "ᵻ": 4
 
 
 
46
  }