Ilyes commited on
Commit
9b91d7a
1 Parent(s): 35b867b

update the model: add "-" to the prediction characters

Browse files

Files changed (4) hide show
  1. README.md +4 -25
  2. pytorch_model.bin +2 -2
  3. pytorch_model_v2.bin +0 -3
  4. vocab.json +1 -1
README.md CHANGED
@@ -44,8 +44,7 @@ ds = load_dataset("common_voice", "fr", split="test", cache_dir="./data/fr")
44
 
45
 
46
 
47
-
48
- chars_to_ignore_regex = '[\\\\,\\\\?\\\\.\\\\!\\\\-\\\\;\\\\:\\\\"\\\\“\\\\%\\\\‘\\\\”\\\\�\\\\‘\\\\’\\\\’\\\\’\\\\‘\\\\…\\\\·\\\\!\\\\ǃ\\\\?\\\\«\\\\‹\\\\»\\\\›“\\\\”\\\\\\\\ʿ\\\\ʾ\\\\„\\\\∞\\\\\\\\|\\\\.\\\\,\\\\;\\\\:\\\\*\\\\—\\\\–\\\\─\\\\―\\\\_\\\\/\\\\:\\\\ː\\\\;\\\\,\\\\=\\\\«\\\\»\\\\→]'
49
  def map_to_array(batch):
50
  speech, _ = torchaudio.load(batch["path"])
51
  batch["speech"] = resampler.forward(speech.squeeze(0)).numpy()
@@ -72,28 +71,8 @@ wer = load_metric("wer")
72
  print(wer.compute(predictions=result["predicted"], references=result["target"]))
73
  ```
74
 
75
- ## Training
76
-
77
- 6% of the Common Voice `train`, `validation` datasets (20K files) were used for training.
78
-
79
- ## Testing
80
-
81
- All the Common Voice `Test` dataset (15763 files) were used for testing.
82
-
83
- Results:
84
-
85
- WER=20.89%
86
-
87
- SER=77.56%
88
-
89
-
90
- ## New Model (v2)
91
-
92
- ~10% of the Common Voice `train`, `validation` datasets (30K files) were used for training.
93
-
94
- Results:
95
-
96
- WER=18.81%
97
 
98
- SER=73.82%
99
 
 
44
 
45
 
46
 
47
+ chars_to_ignore_regex = '[\,\?\.\!\;\:\"\“\%\‘\”\�\‘\’\’\’\‘\…\·\!\ǃ\?\«\‹\»\›“\”\\ʿ\ʾ\„\∞\\|\.\,\;\:\*\—\–\─\―\_\/\:\ː\;\,\=\«\»\→]'
 
48
  def map_to_array(batch):
49
  speech, _ = torchaudio.load(batch["path"])
50
  batch["speech"] = resampler.forward(speech.squeeze(0)).numpy()
71
  print(wer.compute(predictions=result["predicted"], references=result["target"]))
72
  ```
73
 
74
+ ## Results
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
75
 
76
+ WER=18.29%
77
 
78
+ SER=71.44%
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9d5519b860f0eb5200238e304f15f7834fb79be18367eea0138f43f7b67ac495
3
- size 1262097815
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:44440692ceb1c3d778d0b89ec5b662f6485d0c8f51dea99173935e8e616f4bc6
3
+ size 1262101911
pytorch_model_v2.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:58735873545095dd226e8e8064f2b5bcd09bf29f9de74175a7b0cc7b7d777494
3
- size 1262097815
 
 
 
vocab.json CHANGED
@@ -1 +1 @@
1
- {"a": 0, "e": 1, "i": 2, "o": 3, "u": 4, "y": 5, "b": 6, "c": 7, "d": 8, "f": 9, "g": 10, "h": 11, "j": 12, "k": 13, "l": 14, "m": 15, "n": 16, "p": 17, "q": 18, "r": 19, "s": 20, "t": 21, "v": 22, "w": 23, "x": 24, "z": 25, "\u00e0": 26, "\u00e2": 27, "\u00e7": 28, "\u00e8": 29, "\u00e9": 30, "\u00ea": 31, "\u00ee": 32, "\u00f4": 33, "\u00f9": 34, "\u00fb": 35, "|": 36, "'": 37, "<unk>": 38, "<pad>": 39}
1
+ {"a": 0, "e": 1, "i": 2, "o": 3, "u": 4, "y": 5, "b": 6, "c": 7, "d": 8, "f": 9, "g": 10, "h": 11, "j": 12, "k": 13, "l": 14, "m": 15, "n": 16, "p": 17, "q": 18, "r": 19, "s": 20, "t": 21, "v": 22, "w": 23, "x": 24, "z": 25, "à": 26, "â": 27, "ç": 28, "è": 29, "é": 30, "ê": 31, "î": 32, "ô": 33, "ù": 34, "û": 35, "|": 36, "'": 37, "-": 38, "<unk>": 39, "<pad>": 40}