cdactvm commited on
Commit
968793a
·
verified ·
1 Parent(s): 6efc2bb

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +25 -1
app.py CHANGED
@@ -11,6 +11,7 @@ import re
11
 
12
  # Initialize the speech recognition pipeline and transliterator
13
  p1 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-2.0-odia_v1")
 
14
  p2 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-2.0-hindi_v1")
15
  #p3 = pipeline(task="automatic-speech-recognition", model="cdactvm/kannada_w2v-bert_model")
16
  #p4 = pipeline(task="automatic-speech-recognition", model="cdactvm/telugu_w2v-bert_model")
@@ -47,7 +48,26 @@ def transcribe_odiya(speech):
47
  return "Error: ASR returned None"
48
  return text
49
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
50
 
 
51
  def cleanhtml(raw_html):
52
  cleantext = re.sub(r'<.*?>', '', raw_html)
53
  return cleantext
@@ -196,7 +216,11 @@ def sel_lng(lng, mic=None, file=None):
196
  elif lng == "Assamese-LM":
197
  return transcribe_assamese_LM(audio)
198
  elif lng == "Assamese-Model2":
 
 
199
  return transcribe_assamese_model2(audio)
 
 
200
 
201
 
202
  # Function to replace incorrectly spelled words
@@ -423,7 +447,7 @@ demo=gr.Interface(
423
  inputs=[
424
 
425
  #gr.Dropdown(["Hindi","Hindi-trans","Odiya","Odiya-trans","Kannada","Kannada-trans","Telugu","Telugu-trans","Bangala","Bangala-trans"],value="Hindi",label="Select Language"),
426
- gr.Dropdown(["Hindi","Hindi-trans","Odiya","Odiya-trans","Assamese-LM","Assamese-Model2"],value="Hindi",label="Select Language"),
427
  gr.Audio(sources=["microphone","upload"], type="filepath"),
428
  #gr.Audio(sources="upload", type="filepath"),
429
  #"state"
 
11
 
12
  # Initialize the speech recognition pipeline and transliterator
13
  p1 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-2.0-odia_v1")
14
+ odia-model2 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-odia_v1")
15
  p2 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-2.0-hindi_v1")
16
  #p3 = pipeline(task="automatic-speech-recognition", model="cdactvm/kannada_w2v-bert_model")
17
  #p4 = pipeline(task="automatic-speech-recognition", model="cdactvm/telugu_w2v-bert_model")
 
48
  return "Error: ASR returned None"
49
  return text
50
 
51
+ ###################################
52
+ def transcribe_odiya_model2(speech):
53
+ text = odia-model2(speech)["text"]
54
+ if text is None:
55
+ return "Error: ASR returned None"
56
+ return text
57
+
58
+ def transcribe_odiya_eng_model2(speech):
59
+ trn = Transliterator(source='ori', target='eng', build_lookup=True)
60
+ text = odia-model2(speech)["text"]
61
+ if text is None:
62
+ return "Error: ASR returned None"
63
+ sentence = trn.transform(text)
64
+ if sentence is None:
65
+ return "Error: Transliteration returned None"
66
+ replaced_words = replace_words(sentence)
67
+ processed_sentence = process_doubles(replaced_words)
68
+ return process_transcription(processed_sentence)
69
 
70
+ ########################################
71
  def cleanhtml(raw_html):
72
  cleantext = re.sub(r'<.*?>', '', raw_html)
73
  return cleantext
 
216
  elif lng == "Assamese-LM":
217
  return transcribe_assamese_LM(audio)
218
  elif lng == "Assamese-Model2":
219
+ return transcribe_odiya_model2(audio)
220
+ elif lng == "Odia_model2":
221
  return transcribe_assamese_model2(audio)
222
+ elif lng == "Odia_trans_model2":
223
+ return transcribe_odiya_eng_model2(audio)
224
 
225
 
226
  # Function to replace incorrectly spelled words
 
447
  inputs=[
448
 
449
  #gr.Dropdown(["Hindi","Hindi-trans","Odiya","Odiya-trans","Kannada","Kannada-trans","Telugu","Telugu-trans","Bangala","Bangala-trans"],value="Hindi",label="Select Language"),
450
+ gr.Dropdown(["Hindi","Hindi-trans","Odiya","Odiya-trans","Odia_model2","Odia_trans_model2","Assamese-LM","Assamese-Model2"],value="Hindi",label="Select Language"),
451
  gr.Audio(sources=["microphone","upload"], type="filepath"),
452
  #gr.Audio(sources="upload", type="filepath"),
453
  #"state"