I'm a computer programmer and very fascinated by Home Automation. I'm
starting to take a look into the many, many possible options out
there. One thing I'd like to mess with myself is Voice Recognition in
my home. However, I don't want to have to walk up to a microphone on
my computer, blanket every square inch of my home with microphones, or
wear a wireless microphone (or headset).

So, I figured this might be a good place to ask these kinds of
questions. I live in a loft... about 1500 square feet (all open) with
about 15 foot ceilings. The loft is a converted factory near Uptown
Charlotte, NC. What I was thinking about was looking into some options
that would allow me to hook up just 2 or 3 (depending on what's
suggested) GOOD microphones that are small with good range so they can
be hidden in the cross beams that go across my ceiling. I'm trying to
get solid coverage that won't require me to shout or turn off the
television set. Ultimately I'd want the input from all microphones to
be fed into my computer for voice recognition processing.

Any ideas of a good direction for this?

"Matt" wrote in message
om...
I'm a computer programmer and very fascinated by Home Automation. I'm
starting to take a look into the many, many possible options out
there. One thing I'd like to mess with myself is Voice Recognition in
my home. However, I don't want to have to walk up to a microphone on
my computer, blanket every square inch of my home with microphones, or
wear a wireless microphone (or headset).

So, I figured this might be a good place to ask these kinds of
questions. I live in a loft... about 1500 square feet (all open) with
about 15 foot ceilings. The loft is a converted factory near Uptown
Charlotte, NC. What I was thinking about was looking into some options
that would allow me to hook up just 2 or 3 (depending on what's
suggested) GOOD microphones that are small with good range so they can
be hidden in the cross beams that go across my ceiling. I'm trying to
get solid coverage that won't require me to shout or turn off the
television set. Ultimately I'd want the input from all microphones to
be fed into my computer for voice recognition processing.

Any ideas of a good direction for this?

http://www.shure.com/mixers/installed/ams/default.asp

Having been in the Speech (not Voice) Recognition field professionally for
30 years, rolling your own is not trivial, in fact, almost impossible. Get
the Scansoft "Naturally Speaking" or the IBM "Viavoice", both are good
products and relatively inexpensive. Both work out of the box for clean,
well articulated, continuous speech, but with training work a lot better.
Train in the environment you will use them in and most of the relatively
constant background artifacts and echo will be neutralized by the built-in
processing software. One sensitive boundary mic (like the Crown PCC) is
probably better than several, as the others not near you will intro noise
with little signal. Unless you can figure out a way to only turn on the mic
nearest you.

Actually a wireless headset would work great, many use this method already.
This is usually teamed with a phone switch so you can use one headset for
both. There are several very light and good ones on the market.

If you are only going to use it for Command and Control and not for
dictation, I would go with "Dragon Dictate", if you can still find a copy.

The "anti-vox" idea is a very good one since it can be triggered by other
voices coming from anywhere. Use the "active noise control" idea to invert
the phase 180 degrees and cancel out the TV, etc.

comp.speech.users is a good source of info.


"Ben Bradley" wrote in message
...

In rec.audio.pro, (Matt) wrote:
One thing I'd like to mess with myself is Voice Recognition in
my home.

You may need input from the TV, stereo, radio and whatnot into the
software to act as an 'anti-vox' [as the term is used in amateur
radio] so the VR system won't respond to someone on a sitcom saying an
annoying command such as "turn the TV off."