[Tfug] OT? "Alarm/alert" suggestions y3esuzup

[Bexley Hall]

Hi,

I've designed my "network speaker" and have started selecting
production components (current prototype uses COTS parts -- too
big, too expensive, many missing features, etc.).

In addition to spots around the house for "multimedia" use,
I had also planned on locating one in each bathroom.  There,
however, not for "entertainment" but, rather, "communication".

E.g., if you are in the shower, system can tell you "Bob is on
the phone", "Fred is at the front door", etc.

Also, I had originally envisioned adding a "contact closure"
input to the device.  My thinking being that I could have a
pull-chain or "panic button" wired to this to summon help
("Help, I've fallen and I can't get up" TmReg).  The presence
of the loudspeaker function would allow the system to query
the "occupant" to verify that there is a genuine need for
assistance (i.e., don't want to auto-dial '911' only to hear
someone pounding on your front door -- while you're still in
the shower blissfully ignorant of what the system has *done*!).

But, I've decided that a panic-button isn't very useful as it
has to be within arm's reach at all times (where could you
locate it such that it would be accessible if you were at the sink,
in the shower, collapsed on the floor, etc.?)

A pull chain can be marginally more accessible (i.e., at least it
eliminates the "height" requirement!) but is still a poor choice
for something that MUST be reliable.

So, I think I am settling on the notion of making the "alarm"
(alert?) voice-activated (sound activated?).  This removes
physical location from the accessibility issue.

But, bathrooms are inherently noisey environments.  So, trying
to recognize "Help!" is a non-starter.  Likewise, "a loud noise"
could just as easily be a normal "bathroom sound".  Hand claps?
etc.

And, someone who is ill -- perhaps critically -- may not be able
to produce the sorts of sounds (e.g., yell loudly) that you
might be tempted to *think* would be "exceptional enough".

Processing power isn't an issue (within reason).  While the horsepower
that I have available *in* the "network speaker" is tuned to fit its
needs (and very little else!), I can always ship data down the wire for
processing elsewhere -- relying on the network speaker to act as just
the "user I/O"...

So, the question(s) I pose is (are):
- what sort of "indicator" to listen for?
- how to "verify" the intent of that "indicator" (false positives)?
- flaws in this approach, in general?
- alternative approaches??

Thx,
--don

[Apologies if I don't respond promptly.  I'm fighting some health
issues of my own at the moment...]