Update app.py

app.py

@@ -22,22 +22,22 @@ def inference(audio_file, audio_file2):
        predicted_ids = torch.argmax(logits, dim=-1)
      predicted_ids[predicted_ids == -100] = processor.tokenizer.pad_token_id
      transcription1 = processor.tokenizer.batch_decode(predicted_ids)[0]
-    else:
+   else:
      transcription1 = "N/A"
      
-    if audio_file2:
-      input_values = processor(map_to_array(audio_file2.name), return_tensors="pt", padding="longest").input_values  # Batch size 1
-      logits = model(input_values).logits
+   if audio_file2:
+     input_values = processor(map_to_array(audio_file2.name), return_tensors="pt", padding="longest").input_values  # Batch size 1
+     logits = model(input_values).logits
   
-      with torch.no_grad():
-        predicted_ids = torch.argmax(logits, dim=-1)
-      predicted_ids[predicted_ids == -100] = processor.tokenizer.pad_token_id
-      transcription2 = processor.tokenizer.batch_decode(predicted_ids)[0]
-    else :
-      transcription2 = "N/A"
+     with torch.no_grad():
+       predicted_ids = torch.argmax(logits, dim=-1)
+     predicted_ids[predicted_ids == -100] = processor.tokenizer.pad_token_id
+     transcription2 = processor.tokenizer.batch_decode(predicted_ids)[0]
+   else :
+     transcription2 = "N/A"
      
      
-     return transcription1, transcription2
+   return transcription1, transcription2
      
      
 gradio_ui = gr.Interface(