Update app.py

app.py

@@ -10,18 +10,18 @@ import re
 #import torchaudio
 
 # Initialize the speech recognition pipeline and transliterator
-#p1 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-2.0-odia_v1")
-#odia_model2 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-odia_v1")
+odia_model1 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-odia_v1")
+odia_model2 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-odia_v2")
 # p2 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-2.0-hindi_v1")
 # punjaib_modle_30000=pipeline(task="automatic-speech-recognition", model="cdactvm/wav2vec-bert-punjabi-30000-model")
-punjaib_modle_155750=pipeline(task="automatic-speech-recognition", model="cdactvm/wav2vec-bert-punjabi-155750-model")
-punjaib_modle_70000_aug=pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-model-30000-augmented")
+# punjaib_modle_155750=pipeline(task="automatic-speech-recognition", model="cdactvm/wav2vec-bert-punjabi-155750-model")
+# punjaib_modle_70000_aug=pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-bert-model-30000-augmented")
 #p3 = pipeline(task="automatic-speech-recognition", model="cdactvm/kannada_w2v-bert_model")
 #p4 = pipeline(task="automatic-speech-recognition", model="cdactvm/telugu_w2v-bert_model")
 #p5 = pipeline(task="automatic-speech-recognition", model="Sajjo/w2v-bert-2.0-bangala-gpu-CV16.0_v2")
 #p6 = pipeline(task="automatic-speech-recognition", model="cdactvm/hf-open-assames")
 # p7 = pipeline(task="automatic-speech-recognition", model="cdactvm/w2v-assames")
-processor = AutoProcessor.from_pretrained("cdactvm/w2v-assames")
+# processor = AutoProcessor.from_pretrained("cdactvm/w2v-assames")
 vocab_dict = processor.tokenizer.get_vocab()
 sorted_vocab_dict = {k.lower(): v for k, v in sorted(vocab_dict.items(), key=lambda item: item[1])}
 decoder = build_ctcdecoder(
@@ -331,19 +331,30 @@ def transcribe_punjabi_eng_model_155750(speech):
     return sentence
     
 ###########################################
-def transcribe_odiya(speech):
-    text = p1(speech)["text"]
+def transcribe_odiya_model1(speech):
+    text = odia_model1(speech)["text"]
     if text is None:
         return "Error: ASR returned None"
     return text
 
-###################################
 def transcribe_odiya_model2(speech):
     text = odia_model2(speech)["text"]
     if text is None:
         return "Error: ASR returned None"
     return text
 
+def transcribe_odiya_eng_model1(speech):
+    trn = Transliterator(source='ori', target='eng', build_lookup=True)
+    text = odia_model1(speech)["text"]
+    if text is None:
+        return "Error: ASR returned None"
+    sentence = trn.transform(text)
+    if sentence is None:
+        return "Error: Transliteration returned None"
+    replaced_words = replace_words(sentence)
+    processed_sentence = process_doubles(replaced_words)
+    return process_transcription(processed_sentence)
+    
 def transcribe_odiya_eng_model2(speech):
     trn = Transliterator(source='ori', target='eng', build_lookup=True)
     text = odia_model2(speech)["text"]
@@ -552,6 +563,10 @@ def sel_lng(lng, mic=None, file=None):
        return transcribe_assamese_LM(audio)
     elif lng == "Assamese-Model2":
        return transcribe_assamese_model2(audio)
+    elif lng == "Odia_model1":
+       return transcribe_odiya_model1(audio)
+    elif lng == "Odia_trans_model1":
+       return transcribe_odiya_eng_model1(audio)
     elif lng == "Odia_model2":
        return transcribe_odiya_model2(audio)
     elif lng == "Odia_trans_model2":
@@ -603,9 +618,10 @@ demo=gr.Interface(
         #gr.Dropdown(["Hindi","Hindi-trans","Odiya","Odiya-trans","Kannada","Kannada-trans","Telugu","Telugu-trans","Bangala","Bangala-trans"],value="Hindi",label="Select Language"),
         gr.Dropdown([
             # "Hindi","Hindi-trans",
-            # "Odiya","Odiya-trans","Odia_model2","Odia_trans_model2",
+            "Odia_model1","Odiya-trans_model1","Odia_model2","Odia_trans_model2",
             # "Assamese-LM","Assamese-Model2",
-            "Punjabi_Model1","Punjabi_Model1_Trans","Punjabi_Model_aug","Punjabi_Model_aug_Trans"],value="Hindi",label="Select Language"),
+            # "Punjabi_Model1","Punjabi_Model1_Trans","Punjabi_Model_aug","Punjabi_Model_aug_Trans"],value="Hindi",label="Select Language"
+                   ),
         gr.Audio(sources=["microphone","upload"], type="filepath"),
         #gr.Audio(sources="upload", type="filepath"),
         #"state"