boumehdi commited on
Commit
eded877
1 Parent(s): b948b7b

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +17 -2
README.md CHANGED
@@ -17,11 +17,25 @@ model-index:
17
  metrics:
18
  - name: Test WER
19
  type: wer
20
- value: 0.09
21
  ---
22
  # Wav2Vec2-Large-XLSR-53-Moroccan-Darija
23
 
24
- **wav2vec2-large-xlsr-53** fine-tuned on 120 hours of labeled Darija Audios
 
 
 
 
 
 
 
 
 
 
 
 
 
 
25
 
26
  ## Usage
27
 
@@ -58,3 +72,4 @@ print(transcription)
58
  Output: قالت ليا هاد السيد هادا ما كاينش بحالو
59
 
60
  email: souregh@gmail.com
 
 
17
  metrics:
18
  - name: Test WER
19
  type: wer
20
+ value: 0.254919
21
  ---
22
  # Wav2Vec2-Large-XLSR-53-Moroccan-Darija
23
 
24
+ **wav2vec2-large-xlsr-53** fine-tuned on 27 hours (27 people) of labeled Darija Audios.
25
+
26
+ # Old model vs new model
27
+
28
+ Old Model:
29
+ - The model contains numerous incorrect transcriptions as input
30
+ - There are multiple transcribers.
31
+ - The audio database is not organized (by gender, age, regions ..).
32
+ - Wrong wer rate
33
+
34
+ New Model:
35
+ - Transcriptions are now performed by a single individual.
36
+ - Each hour of audio is pronounced by one person.
37
+ - Fine-tuning is ongoing 24/7 to enhance accuracy, and we are consistently adding more data to the model every day.
38
+ - Correct Wer rate
39
 
40
  ## Usage
41
 
 
72
  Output: قالت ليا هاد السيد هادا ما كاينش بحالو
73
 
74
  email: souregh@gmail.com
75
+ BOUMEHDI Ahmed