Return to the archive index

Re: Audio Input/Output in Linux

for simple input, not really input but just commands, you could use the
classic style of voice input/recognition, which just infinitely clips the
sound to make square waves then compare to a database of prerecorded ones,
have a few different recordings of each command and you get pretty good
recognition, so you could have simple things like
read mail
next
save
close
news
next
save
quit
skip
mark
etc

it doesn't require tons of processing power either. In an issue of Circuit
Cellar Ink a bit ago they made a hardware device costing maybe $10 in
actual parts that could recognize 16 commands, or less if you duplicate
for better recognition, that just turned on leds, but with a few of these
it could output to the parallel port and get a lot more commands.

so type when you need real input but when just walking around, this could
probably work quite well. as who wants to MUTTER (like a crazy person) the
great american novel while standing in line at the supermarket (looking
like a BORG) anyway. or at least the muttering part.

Bryan

In article <m3vhm8v29d.fsf@zen.binary.net.au>,
David Maslen  <david@binary.net.au> wrote:
>Well you may remeber I've been putting together a keyglove. Well it's
>not quite finished, but I don't find my prototype very promising. I
>don't think I'll be happy with this type of input in the long term,
>probably when I want to type I'll use a keyboard.
>
>Has anyone had any success using linux voice recognition software to
>run their linux wearable. Xemacs-speech is probably the best way to
>get audio output, but how about audio input for sometype of keyword
>activation?
>
>-- 
>Binary Bar - Australia's first free access internet bar/cafe/gallery.
>243 Brunswick Street, Fitzroy, Melbourne, Australia. 3pm - 1am
>http://www.binary.net.au/
>