Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -18,6 +18,7 @@ p1 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-2.0-o
|
|
| 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 |
|
|
@@ -47,6 +48,13 @@ def transcribe_telugu(speech):
|
|
| 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"]
|
|
@@ -58,6 +66,18 @@ def transcribe_odiya_eng(speech):
|
|
| 58 |
replaced_words = replace_words(sentence)
|
| 59 |
processed_sentence = process_doubles(replaced_words)
|
| 60 |
return process_transcription(processed_sentence)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 61 |
|
| 62 |
def transcribe_hin_eng(speech):
|
| 63 |
trn = Transliterator(source='hin', target='eng', build_lookup=True)
|
|
@@ -136,6 +156,10 @@ def sel_lng(lng, mic=None, file=None):
|
|
| 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):
|
|
|
|
| 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 |
+
p5 = pipeline(task="automatic-speech-recognition", model="Sajjo/w2v-bert-2.0-bangala-gpu-CV16.0_v2")
|
| 22 |
os.system('git clone https://github.com/irshadbhat/indic-trans.git')
|
| 23 |
os.system('pip install ./indic-trans/.')
|
| 24 |
|
|
|
|
| 48 |
if text is None:
|
| 49 |
return "Error: ASR returned None"
|
| 50 |
return text
|
| 51 |
+
|
| 52 |
+
def transcribe_bangala(speech):
|
| 53 |
+
text = p5(speech)["text"]
|
| 54 |
+
if text is None:
|
| 55 |
+
return "Error: ASR returned None"
|
| 56 |
+
return text
|
| 57 |
+
|
| 58 |
def transcribe_odiya_eng(speech):
|
| 59 |
trn = Transliterator(source='ori', target='eng', build_lookup=True)
|
| 60 |
text = p1(speech)["text"]
|
|
|
|
| 66 |
replaced_words = replace_words(sentence)
|
| 67 |
processed_sentence = process_doubles(replaced_words)
|
| 68 |
return process_transcription(processed_sentence)
|
| 69 |
+
|
| 70 |
+
def transcribe_ban_eng(speech):
|
| 71 |
+
trn = Transliterator(source='ori', target='eng', build_lookup=True)
|
| 72 |
+
text = p5(speech)["text"]
|
| 73 |
+
if text is None:
|
| 74 |
+
return "Error: ASR returned None"
|
| 75 |
+
sentence = trn.transform(text)
|
| 76 |
+
if sentence is None:
|
| 77 |
+
return "Error: Transliteration returned None"
|
| 78 |
+
replaced_words = replace_words(sentence)
|
| 79 |
+
processed_sentence = process_doubles(replaced_words)
|
| 80 |
+
return process_transcription(processed_sentence)
|
| 81 |
|
| 82 |
def transcribe_hin_eng(speech):
|
| 83 |
trn = Transliterator(source='hin', target='eng', build_lookup=True)
|
|
|
|
| 156 |
return transcribe_tel_eng(audio)
|
| 157 |
elif lng == "Telugu":
|
| 158 |
return transcribe_telugu(audio)
|
| 159 |
+
elif lng == "Bangala-trans":
|
| 160 |
+
return transcribe_ban_eng(audio)
|
| 161 |
+
elif lng == "Bangala":
|
| 162 |
+
return transcribe_bangala(audio)
|
| 163 |
|
| 164 |
# Function to replace incorrectly spelled words
|
| 165 |
def replace_words(sentence):
|