You can download the result - here.
If you don't know what TVSA3 does, the docs from the original package are here.
In summary, It's an old program written by some random paranoid guy that analyzes wav files of people speaking, and marks points where they are under higher stress than normal. Sometimes the stress is caused by the speaker lying, or not telling the whole truth, other times it might be completely random.
It takes in wav files that are 16 bit, mono, 11.025 khz, and spits them out with beeps inserted where the stress exceeds a specified percentage of the maximum for that sample.
On a completely unrelated topic, have you noticed that the whitehouse web site has an archive with audio files of all of Bush's radio addresses?