Update app.py

app.py

@@ -15,7 +15,7 @@ def read_file_and_process(wav_file):
     resampler(wav_file, filename_16k)
     speech, _ = sf.read(filename_16k)
     inputs = processor(speech, sampling_rate=16_000, return_tensors="pt", padding=True)
-    
+
     return inputs
 
 
@@ -33,18 +33,29 @@ def parse_transcription(logits):
     return transcription
 
 
+# def parse(wav_file):
+#     input_values = read_file_and_process(wav_file)
+#     with torch.no_grad():
+#         logits = model(**input_values).logits
+#     sentence = parse_transcription(logits)
+#     check_spell = spell_checker.check(sentence)
+#     if check_spell[0] is False:
+#       corrected = check_spell[1]
+#     else:
+#       corrected = sentence
+#     return corrected
+
 def parse(wav_file):
     input_values = read_file_and_process(wav_file)
     with torch.no_grad():
         logits = model(**input_values).logits
-    sentence = parse_transcription(logits)
-    check_spell = spell_checker.check(sentence)
-    if check_spell[0] is False:
-      corrected = check_spell[1] 
-    else:
-      corrected = sentence
-    return corrected
-
+    # sentence = parse_transcription(logits)
+    check_spell = spell_checker.check(parse_transcription(logits))
+    # if check_spell[0] is False:
+    #   corrected = check_spell[1]
+    # else:
+    #   corrected = sentence
+    return spell_checker.check(parse_transcription(logits))[1] if spell_checker.check(parse_transcription(logits))[0] is False else parse_transcription(logits)
 
 model_id = "jonatasgrosman/wav2vec2-large-xlsr-53-persian"
 processor = Wav2Vec2Processor.from_pretrained(model_id)