cdactvm commited on
Commit
ee6e935
·
verified ·
1 Parent(s): d4e2ce0

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +46 -2
app.py CHANGED
@@ -16,6 +16,8 @@ pipe = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-2.0
16
 
17
  p1 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-2.0-odia_v1")
18
  p2 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-2.0-hindi_v1")
 
 
19
  os.system('git clone https://github.com/irshadbhat/indic-trans.git')
20
  os.system('pip install ./indic-trans/.')
21
 
@@ -35,7 +37,16 @@ def transcribe_hindi(speech):
35
  if text is None:
36
  return "Error: ASR returned None"
37
  return text
38
-
 
 
 
 
 
 
 
 
 
39
  def transcribe_odiya_eng(speech):
40
  trn = Transliterator(source='ori', target='eng', build_lookup=True)
41
  text = p1(speech)["text"]
@@ -60,6 +71,31 @@ def transcribe_hin_eng(speech):
60
  processed_sentence = process_doubles(replaced_words)
61
  return process_transcription(processed_sentence)
62
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
63
  def process_transcription(input_sentence):
64
  word_to_code_map = {}
65
  code_to_word_map = {}
@@ -92,6 +128,14 @@ def sel_lng(lng, mic=None, file=None):
92
  return transcribe_hin_eng(audio)
93
  elif lng == "Hindi":
94
  return transcribe_hindi(audio)
 
 
 
 
 
 
 
 
95
 
96
  # Function to replace incorrectly spelled words
97
  def replace_words(sentence):
@@ -316,7 +360,7 @@ demo=gr.Interface(
316
 
317
  inputs=[
318
 
319
- gr.Dropdown(["Hindi","Hindi-trans","Odiya","Odiya-trans"],value="Hindi",label="Select Language"),
320
  gr.Audio(sources=["microphone","upload"], type="filepath"),
321
  #gr.Audio(sources="upload", type="filepath"),
322
  #"state"
 
16
 
17
  p1 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-2.0-odia_v1")
18
  p2 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-2.0-hindi_v1")
19
+ p3 = pipeline(task="automatic-speech-recognition", model="cdactvm/kannada_w2v-bert_model")
20
+ p4 = pipeline(task="automatic-speech-recognition", model="cdactvm/telugu_w2v-bert_model")
21
  os.system('git clone https://github.com/irshadbhat/indic-trans.git')
22
  os.system('pip install ./indic-trans/.')
23
 
 
37
  if text is None:
38
  return "Error: ASR returned None"
39
  return text
40
+ def transcribe_kannada(speech):
41
+ text = p3(speech)["text"]
42
+ if text is None:
43
+ return "Error: ASR returned None"
44
+ return text
45
+ def transcribe_telugu(speech):
46
+ text = p4(speech)["text"]
47
+ if text is None:
48
+ return "Error: ASR returned None"
49
+ return text
50
  def transcribe_odiya_eng(speech):
51
  trn = Transliterator(source='ori', target='eng', build_lookup=True)
52
  text = p1(speech)["text"]
 
71
  processed_sentence = process_doubles(replaced_words)
72
  return process_transcription(processed_sentence)
73
 
74
+ def transcribe_kan_eng(speech):
75
+ trn = Transliterator(source='kan', target='eng', build_lookup=True)
76
+ text = p3(speech)["text"]
77
+ if text is None:
78
+ return "Error: ASR returned None"
79
+ sentence = trn.transform(text)
80
+ if sentence is None:
81
+ return "Error: Transliteration returned None"
82
+ replaced_words = replace_words(sentence)
83
+ processed_sentence = process_doubles(replaced_words)
84
+ return process_transcription(processed_sentence)
85
+
86
+ def transcribe_tel_eng(speech):
87
+ trn = Transliterator(source='tel', target='eng', build_lookup=True)
88
+ text = p4(speech)["text"]
89
+ if text is None:
90
+ return "Error: ASR returned None"
91
+ sentence = trn.transform(text)
92
+ if sentence is None:
93
+ return "Error: Transliteration returned None"
94
+ replaced_words = replace_words(sentence)
95
+ processed_sentence = process_doubles(replaced_words)
96
+ return process_transcription(processed_sentence)
97
+
98
+
99
  def process_transcription(input_sentence):
100
  word_to_code_map = {}
101
  code_to_word_map = {}
 
128
  return transcribe_hin_eng(audio)
129
  elif lng == "Hindi":
130
  return transcribe_hindi(audio)
131
+ elif lng == "Kannada-trans":
132
+ return transcribe_kan_eng(audio)
133
+ elif lng == "Kannada":
134
+ return transcribe_kannada(audio)
135
+ elif lng == "Telugu-trans":
136
+ return transcribe_tel_eng(audio)
137
+ elif lng == "Telugu":
138
+ return transcribe_telugu(audio)
139
 
140
  # Function to replace incorrectly spelled words
141
  def replace_words(sentence):
 
360
 
361
  inputs=[
362
 
363
+ gr.Dropdown(["Hindi","Hindi-trans","Odiya","Odiya-trans","Kannada","Kannada-trans","Telugu","Telugu-trans"],value="Hindi",label="Select Language"),
364
  gr.Audio(sources=["microphone","upload"], type="filepath"),
365
  #gr.Audio(sources="upload", type="filepath"),
366
  #"state"