So I wrote a rough pass for my OneNote PowerToy: it synchronizes voice recordings on a SmartPhone / PocketPC onto a desktop and imports them as links into a OneNote page. After spending some time with a hex editor, I backed off the idea of trying to do a binary edit of the OneNote file to embed the links as native OneNote audio embeds (even though, in fact, OneNote embeds a link to a file, so it ought to be possible to do the binary edit, but not for a Sunday morning project).

Unfortunately, using a file link results in an annoying “Are you sure you want to open this potentially unsafe file?” dialog every time you click on it. I'm pretty sure that this annoyance will drop my PowerToy back at least to the middle of the pack. So I thought, “Well, what can I do that would be slick?” and my thought was “Automatic voice recognition of the voice notes, at least to get a running shot at search / browsability.”

It turns out that voice recognition of a file could hardly be easier: it's literally a 10-line initialization block and then an event-handler for the results. When tried on a .wav file generated from my noise-cancelling Plantronics headset the results are definitely adequate, but the results on the low-fidelity recordings from a phone are so poor that I'm not even going to include the capability in the PowerToy.

Oh, hey, today's our 1-year anniversary of moving to Hawaii. Rockin'. Had a great time at Honaunau yesterday freediving. My freediving has taken a big step in the past couple months -- I'm now really comfortable for about 1:30 on the bottom, which is great, because the reef fish change from being scared to being curious at about 1:00. My goal is 2:00 on the bottom, with :30 of swimming up and down time. I'm also getting more and more confident about taking a shot at the 100' club...