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
929ee00
·
verified ·
1 Parent(s): e4cd08b

Upload tokenizer

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