sha1779 commited on
Commit
49da3ad
·
verified ·
1 Parent(s): 676724b
Files changed (1) hide show
  1. README.md +11 -3
README.md CHANGED
@@ -4,11 +4,14 @@ language:
4
  - bn
5
  base_model:
6
  - openai/whisper-small
 
7
  pipeline_tag: automatic-speech-recognition
8
 
9
  ---
10
- ```py
11
 
 
 
12
  import os
13
  import librosa
14
  import torch, torchaudio
@@ -22,7 +25,7 @@ processor = WhisperProcessor.from_pretrained(model_path_)
22
  model = WhisperForConditionalGeneration.from_pretrained(model_path_).to(device)
23
  model.config.forced_decoder_ids = processor.get_decoder_prompt_ids(language="bengali", task="transcribe")
24
 
25
- mp3_path = "/kaggle/input/barishal-data/valid_barishal (104).wav"
26
  speech_array, sampling_rate = librosa.load(mp3_path, sr=16000)
27
 
28
  speech_array = librosa.resample(np.asarray(speech_array), orig_sr=sampling_rate, target_sr=16000)
@@ -34,5 +37,10 @@ transcription = processor.decode(predicted_ids, skip_special_tokens=True)
34
 
35
  print(transcription)
36
 
 
 
 
 
 
 
37
 
38
- ```
 
4
  - bn
5
  base_model:
6
  - openai/whisper-small
7
+
8
  pipeline_tag: automatic-speech-recognition
9
 
10
  ---
11
+ BengaliRegionalASR trained on bengali regional dialact dataset.
12
 
13
+ # Try the model
14
+ ```py
15
  import os
16
  import librosa
17
  import torch, torchaudio
 
25
  model = WhisperForConditionalGeneration.from_pretrained(model_path_).to(device)
26
  model.config.forced_decoder_ids = processor.get_decoder_prompt_ids(language="bengali", task="transcribe")
27
 
28
+ mp3_path = "https://huggingface.co/sha1779/BengaliRegionalASR/resolve/main/Mp3/common_voice_bn_31617644.mp3"
29
  speech_array, sampling_rate = librosa.load(mp3_path, sr=16000)
30
 
31
  speech_array = librosa.resample(np.asarray(speech_array), orig_sr=sampling_rate, target_sr=16000)
 
37
 
38
  print(transcription)
39
 
40
+ ```
41
+
42
+ # Evaluation
43
+ Word Error Rate 0.65 %
44
+
45
+
46