Update app.py

app.py

@@ -15,8 +15,25 @@ def transcribe_audio(audio_file):
 
 # Function to correct grammar in text
 def correct_text(raw_text):
-    corrected = correction_pipeline(raw_text, max_length=200, num_return_sequences=1)[0]["generated_text"]
-    return corrected
+    try:
+        # Chunk the text into smaller pieces if needed (to handle token limits)
+        text_chunks = chunk_text(raw_text, max_tokens=2000)
+        corrected_chunks = []
+
+        for chunk in text_chunks:
+            # Provide clear instructions to the model
+            prompt = f"Correct the following text for grammar and punctuation without changing its meaning: {chunk}"
+            corrected = correction_pipeline(prompt, max_length=2048, num_return_sequences=1)[0]["generated_text"]
+
+            # Add only the corrected part
+            corrected_chunks.append(corrected.strip())
+
+        # Combine corrected chunks
+        final_corrected_text = " ".join(corrected_chunks).strip()
+        return final_corrected_text
+
+    except Exception as e:
+        return f"Error in correction: {str(e)}"
 
 # Function to process the pipeline
 def process_pipeline(audio_file):