This is really cool! The actual experience of standing across the room and whistling at my computer --- and seeing it recognize my call --- is exciting. However, it is really annoying Beatrice at the moment, so I'm going to have to tune this more some other time.
Dave Long writes: > It works off of the zero-crossings from /dev/audio, and is tuned for > what my mic. produces. Enable the debugging output to tune the band > bucketing calculation. > > I tried much fancier approaches, with mexican hats and derivatives > thereof, but none of them seemed to work any better. FWIW /dev/audio normally produces mulaw logarithmic-scale output, which is probably better than linear output for visual analysis, but will pretty thoroughly foil wavelet analysis. Did you correct for that? (I guess I should try it.) I'd expect good results from zero-crossing analysis unless there was (any) high-frequency noise or (large p-t-p amplitude) low-frequency noise. It seems to be relatively easy to get false positives from fan noise on my laptop, unfortunately. What's ups[9] for? -- <[EMAIL PROTECTED]> Kragen Sitaker <http://www.pobox.com/~kragen/> A well designed system must take people into account. . . . It's hard to build a system that provides strong authentication on top of systems that can be penetrated by knowing someone's mother's maiden name. -- Schneier