Hugging Face's logo

RuudVelo
/
XLSR-Wav2Vec2-Maltese-1

Automatic Speech Recognition
Transformers
PyTorch
JAX
Maltese
wav2vec2
audio
speech
xlsr-fine-tuning-week
Eval Results (legacy)
Model card Files
xet
 Community
1
patrickvonplaten commited on
Commit
abfb46c
·
1 Parent(s): 2a081f7

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +3 -3
README.md CHANGED
@@ -19,7 +19,7 @@ model-index:
19
  metrics:
20
  - name: Test WER
21
  type: wer
22
- value: 29.87
23
  ---
24
 
25
 
@@ -38,7 +38,7 @@ import sys
38
 
39
  model_name = "RuudVelo/XLSR-Wav2Vec2-Maltese-1"
40
  device = "cuda"
41
- chars_to_ignore_regex = '[\,\?\.\!\-\;\:\"\“\%\‘\”\�]'
42
 
43
  model = Wav2Vec2ForCTC.from_pretrained(model_name).to(device)
44
  processor = Wav2Vec2Processor.from_pretrained(model_name)
@@ -73,4 +73,4 @@ wer = load_metric("wer")
73
  print(wer.compute(predictions=result["predicted"], references=result["target"]))
74
  ```
75
 
76
- **Result**: 29.87 %
 
19
  metrics:
20
  - name: Test WER
21
  type: wer
22
+ value: 30.0
23
  ---
24
 
25
 
 
38
 
39
  model_name = "RuudVelo/XLSR-Wav2Vec2-Maltese-1"
40
  device = "cuda"
41
+ chars_to_ignore_regex = '[\\,\\?\\.\\!\\-\\;\\:\\"\\“\\%\\‘\\”\\�]'
42
 
43
  model = Wav2Vec2ForCTC.from_pretrained(model_name).to(device)
44
  processor = Wav2Vec2Processor.from_pretrained(model_name)
 
73
  print(wer.compute(predictions=result["predicted"], references=result["target"]))
74
  ```
75
 
76
+ **Result**: 30.0 %