Update app.py

app.py

@@ -114,6 +114,40 @@ def process_video(video_file, language="hindi", progress=gr.Progress()):
         
         print(f"Transcription complete: {len(transcription_result['chunks'])} segments")
         
+        # Post-process: Split overly long chunks for better readability
+        processed_chunks = []
+        max_chunk_duration = 5.0  # Maximum 5 seconds per subtitle
+        max_words_per_chunk = 15  # Maximum words per subtitle
+        
+        for chunk in transcription_result['chunks']:
+            start, end = chunk['timestamp']
+            text = chunk['text'].strip()
+            duration = end - start
+            words = text.split()
+            
+            # If chunk is short enough, keep as is
+            if duration <= max_chunk_duration and len(words) <= max_words_per_chunk:
+                processed_chunks.append(chunk)
+            else:
+                # Split long chunk into smaller pieces
+                words_per_split = max(5, len(words) // max(2, int(duration / max_chunk_duration)))
+                time_per_word = duration / len(words)
+                
+                for i in range(0, len(words), words_per_split):
+                    split_words = words[i:i + words_per_split]
+                    split_start = start + (i * time_per_word)
+                    split_end = start + ((i + len(split_words)) * time_per_word)
+                    
+                    processed_chunks.append({
+                        'timestamp': (split_start, split_end),
+                        'text': ' '.join(split_words)
+                    })
+        
+        print(f"After splitting long chunks: {len(processed_chunks)} segments")
+        
+        # Use processed chunks instead of original
+        transcription_result['chunks'] = processed_chunks
+        
         # Create SRT in output directory (persistent)
         progress(0.6, desc="Creating SRT...")
         srt_path = os.path.join(output_dir, "subtitles.srt")