drunkenbatman Podcast (sorta)
Every once in awhile I get something weird in my inbox that just makes plugging away at these x86 questions worthwhile. Earlier a guy in the Canadian military (really, they do have one) filled me in on his quest to get pictures of The Cow with a fighter jet, and then I got this beauty:
For fun, last night I cobbled together a Perl script to take some UTI HTML, cleanse it, split it up into question-answer pairs, have them read by the OS X `say` command (questions and answers having different voices), have them MP3-ized by LAME, have them all concatenated by another perl script (`mp3cat`, part of `mp3cut`, yes c*u*t, in CPAN), and finally have the concatenated file MP3-ized.I ran your interview through it, optimized the MP3 and added metadata in iTunes, handcrafted a podcast RSS feed and uploaded it.
What a cool hack, and while the .mp3 is a little rough it's pretty damn amusing. I'm just grateful he didn't make me have a female voice. It's almost 2.5 hours in duration, and if you really want a laugh you can fast-forward to about 2:00:20. You can get this in two ways:
- Jesper was kind enough to create a podcast for those who have it setup, which you can get at his site.
- If you'd like to save Jesper some bandwidth, you can download the MP3 from DrunkenBlog [16.5 MB].
FYI, if you haven't seen them, Jesper's two graphs [ graph 1 | graph 2 ] poking fun at the length of the 'interview' (and I use the term loosely) we did are priceless.
Comments (10)
Posted by: drunkenbatman at June 24, 2005 12:28 AM
note: graph 1 and graph 2 point to the same image.
note: fixed. thanks for the catch.
Posted by: Gen Kanai at June 24, 2005 12:55 AM
You might want to submit the podcast to Odeo.com - the new podcasting service.
Posted by: abram at June 24, 2005 01:03 AM
You might want to submit the podcast to Odeo.com - the new podcasting service.
Wait until someone gets two children to do it!
Posted by: Oliver at June 24, 2005 02:46 AM
Nice graphs. Out of curiosity which app did you use to make them?
Posted by: Jay Contonio at June 24, 2005 11:48 AM
Yeah those graphs are definitely pretty.
Posted by: Ciphex at June 24, 2005 07:21 PM
yeah. so it takes a real nerd to come up with a hack like that for fun. i think it kickass! db... i've got plenty of Bibles but never met a Bible salesman.
the voice he gave you... well... does it fit the business?
Posted by: drunkenbatman at June 24, 2005 07:25 PM
Wait until someone gets two children to do it!
'abram', consider yourself banned from the daily show. children shouldn't be anywhere near that interview.
the voice he gave you... well... does it fit the business?
I probably sound more like 'Victoria'. :( I swear, I think he intentionally chose the slowest and most depressing one.
Posted by: Jesper at June 26, 2005 07:52 PM
Thanks, everyone!
I chose voices by random - I didn't even listen once! While slow and depressing could be fitting epithets perchance, I didn't know at the time.
The charts were made by making a Keynote document, scaling down the slide size to 500x400, set the background to a smooth blue gradient and the font to Myriad Pro, and I then, also, somewhere in the heat of it all, made charts, quite simply. Keynote certainly isn't built to make charts, but they look very sharp. (Keynote's roughness is also why the second one has footnotes - the "series" names don't wrap!)
I'm considering making the perl script available once I make it, well, *suck less* (for a heroic piece of glue between amazing open-source thingamabob-whatchamacallits, it's really quite sloppy), but I intentionally wrote up the other parts of the puzzle in somewhat minute detail to make the whole deal easier for brave souls looking to duplicate the effort in some way.
And lastly, for what it's worth, infinite thanks to DB! Bandwidth isn't cheap, and it's part of what kept me from going 'public' on the UTI site itself with this. Let me know what you think of this (wootest at gmail dot com), including improvements that could be made to the actual audio processing, better "parts" to use or simply ideas for text processing before audio conversion.
Posted by: Matthias Lilke at June 27, 2005 03:50 AM
Wow! I had to do a stint in the German Air Force (yes, we too have one ;-) but I never got around to do things like that. Again, wow!! I sure wouldn't mind if someone like that would share his intellectual property... Sounds like something for automator.








note: graph 1 and graph 2 point to the same image.