Jury finds George Zimmerman not guilty

[Igel]

255. It was crap data.

It passed through too many filters whose properties weren't known.

It wasn't loud. You'd have to amplify it. All the noise in the signal gets amplified. You want details. You get noise. I've transcribed 911 recordings. Sometimes all you get is noise and it's hard to tell if there's even a signal.

Assume you can get around that. Having a long enough sample would help. They didn't. Short sample, short amoung of crap. Assume you can get around *that*.

The mode of phonation was non-standard. That means you'd need a sample of that kind of phonation, and you'd need it from both of them. TM was in no position to provide a sample.

Digital blueprints look at details in formant structure, details in coarticulation, idiosyncrasies in modal phonation. No formants. No coarticulation. Modal phonation, to be sure, but the spectral tilt would be a killer (so to speak).

Sorry. Did too much linguistics in grad school, hung out with far too many linguists, and then had the lack of common sense to marry a phonologist/phonetician who spent a decade doing fairly low level acoustic analysis as a post-doc then as faculty at a Tier 1 university.