I find the test a bit unreliable as well. Only someone well trained in giving speeches would read something twice with the exact same tone, diction and pace.

If he were to record the paragraph and play it over speakers or use that same recording twice, then we could see which system is more accurate.