MehdiHosseiniMoghadam commited on
Commit
5dc5731
1 Parent(s): 9c268f9

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +5 -5
README.md CHANGED
@@ -21,7 +21,7 @@ model-index:
21
  metrics:
22
  - name: Test WER
23
  type: wer
24
- value: 26.156584
25
  ---
26
 
27
  # wav2vec2-large-xlsr-53-German
@@ -96,7 +96,7 @@ from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
96
 
97
  import re
98
 
99
- test_dataset = load_dataset("common_voice", "de", split="test[:10%]")
100
 
101
  wer = load_metric("wer")
102
 
@@ -106,7 +106,7 @@ model = Wav2Vec2ForCTC.from_pretrained("MehdiHosseiniMoghadam/wav2vec2-large-xls
106
 
107
  model.to("cuda")
108
 
109
- chars_to_ignore_regex = '[\\,\\?\\.\\!\\-\\;\\:\\"\\“\\%\\‘\\”\\�]'
110
 
111
  resampler = torchaudio.transforms.Resample(48_000, 16_000)
112
 
@@ -166,7 +166,7 @@ print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"],
166
 
167
  ```
168
 
169
- **Test Result**: 26.156584 %
170
 
171
  ## Training
172
 
@@ -174,4 +174,4 @@ print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"],
174
 
175
  ## Testing
176
 
177
- 10% of the Common Voice `Test` dataset were used for training.
21
  metrics:
22
  - name: Test WER
23
  type: wer
24
+ value: 25.284593
25
  ---
26
 
27
  # wav2vec2-large-xlsr-53-German
96
 
97
  import re
98
 
99
+ test_dataset = load_dataset("common_voice", "de", split="test[:15%]")
100
 
101
  wer = load_metric("wer")
102
 
106
 
107
  model.to("cuda")
108
 
109
+ chars_to_ignore_regex = '[\,\?\.\!\-\;\:\"\“\%\‘\”\�]'
110
 
111
  resampler = torchaudio.transforms.Resample(48_000, 16_000)
112
 
166
 
167
  ```
168
 
169
+ **Test Result**: 25.284593 %
170
 
171
  ## Training
172
 
174
 
175
  ## Testing
176
 
177
+ 15% of the Common Voice `Test` dataset were used for training.