OK, so yer lie detector. It’s been something of a blogosphere hit. And in the comments, we have Nigel, who appears to know something about acoustic signal processing – in the sense of “makes speech recognition systems for Eurofighters”.
It seems that rather than being a signal at a frequency between 8 and 12Hz, the signal you’re interested in is a signal, of that frequency, modulated onto the main signal. So in fact, you could theoretically detect it through a telephone call. I was wrong.
However, that isn’t what Nemesysco’s patent claims, and they vigorously deny that what they are doing is voice stress analysis. It’s not the pitch of any such signal that is discussed in the patent, either; it’s the change in the numbers of thorns and plateaus.
Our acoustic expert says that this could be a way of measuring the signals required for classical VSA, just not a very good one; and anyway, he argues that VSA itself is useless, even if it was VSA they were promising to conduct. And, of course, they deny that this is their methodology. Further, VSA gives only one measurement, one of vaguely-defined stress – not the nine or so Nemesysco claim to get out of this.
Meanwhile, someone who makes the same spelling mistakes as Amir Liberman does showed up in comments to claim there was more, secret technology involved that they hadn’t actually patented. Interestingly, he showed up from the same network as Nemesysco’s Web site. The same network was also the source of a Wikipedia article which got deleted for advertising, in which Nemesysco claimed that their method uses 129 different measurements and isn’t anything like VSA. No, sir. And there weren’t 129 different metrics in their patent…