They claim the "Right Quick" EVP would have been "below the audible range" (below 16.4 Hertz, that is, as the subtitles points out), and that it would have become audible through the recorder.

Wow, just wow…

First of all, there isn't any effect by which a recorder would just so shift infrasound into the audible range. At best, you could make your speaker pop and click if the signal is too loud (ultrasound would be a different matter, check my "Creating EVP through Ultrasound" clip http://www.youtube.com/watch?v=HuR_O3HGD4w).

But even if there was such a distortion, there is no possible way that a signal in the 0Hz-16Hz range could contain anything that even remotely resembled a voice. The change rate of such a signal would be way too low. For comparison, in telephony, the frequency band for reproducing voices ranges from approximately 300Hz to 3400Hz. If voices could be stored in the 0Hz-16Hz range, such an infrasound signal could be digitally captured with a sufficient sampling rate of 32 samples per second. With, say, 8 bit per sample this would yield a voice recording at a bitrate of 256bps (without even using compression)! For comparison, the voice codecs of the GSM phone standard compress voice signals at 6500bps (and the resulting sound is already pretty bad).

How comes the phone companies don't know about this fantastic "Ghost Lab" technology that would help them save no less than 95% of their bandwidth?

Anyway, that had to be the silliest piece of pseudo science I have come across so far.