Adjust chunk and stride lengths in the transcription pipeline in app.py to optimize processing efficiency and improve transcription accuracy.

app.py

@@ -42,8 +42,8 @@ transcriber = pipeline(
     tokenizer=processor.tokenizer,
     feature_extractor=processor.feature_extractor,
     device=-1,  # CPU
-    chunk_length_s=60,
-    stride_length_s=(5, 15),
+    chunk_length_s=20,
+    stride_length_s=(5, 5),
 )