Update app.py

app.py

@@ -11,6 +11,7 @@ import evaluate
 from datasets import load_dataset, Audio, disable_caching, set_caching_enabled
 import gradio as gr
 import torch
+import re
 
 set_caching_enabled(False)
 disable_caching()
@@ -28,7 +29,16 @@ model = WhisperForConditionalGeneration.from_pretrained("mskov/whisper-small-esc
 # Evaluate the model
 # model.eval()
 #print("model.eval ", model.eval())
+
+
+# Remove brackets and extra spaces
+
+    
 def map_to_pred(batch):
+    cleaned_transcription = re.sub(r'\[[^\]]+\]', '', batch).strip()
+    cleaned_transcription = preprocess_transcription(batch['sentence'])
+    normalized_transcription = processor.tokenizer._normalize(cleaned_transcription)
+
     audio = batch["audio"]
     input_features = processor(audio["array"], sampling_rate=audio["sampling_rate"], return_tensors="pt").input_features
     batch["reference"] = processor.tokenizer._normalize(batch['sentence'])