Depends on methodics used in testing i guess. After all, part of pump noise is made by air bubbles hitting impeller in not completely bled loop, so depending on if in testing loop has been fully bled or not (sometimes takes long time in some LC setups), noise might differ in tests/daily use.