I have a crown pzm 6R in each room. It's mounted on the wall I'm most likely to face when in the room. In my recording studio, it's on the wall behind the desk, in the living room, it's on the wall behind the tv, in the bedroom, it's on the wall I face when laying in bed, etc... the distance from the ceiling depends on the architectural features of the room (in the living room, there is a 45* angle slope from wall to ceiling which creates my best response). These mics were originally designed with theatrical productions in mind, so they are paintable (5 minutes to tape prep and a shot of spraypaint) so they are painted to compliment the color of the wall.
Each mic runs to a seperate channel of an ooooooooooooooold dbx unit with interchangeble modules (sorry, can't remember the model number). The threshold depends on the acoustic properties of the room. The living room is very bright and therefore has a higher threshold than the bedroom where lots of soft things absorb the sound.
The outputs of the dbx unit each go to a different channel on an old alesis 12r rackmount mixer. the channel strip eq's only have 2 bands, and I clock the hi at about 2 o'clock and the lo at about 11 o'clock with minor adjustments to match the room's acoustical properties.
The output of the 12r uses a 1/4" TRS to mini stereo cable and goes into the sound card input. The sound card output goes to a JVC stereo via a mini stereo to (2)RCA male adapeter. The stereo's speaker outputs go to a sima speaker selector and spread to the small optimus 40wt bookshelf speaker's that serve as Lexxie's voice in each room as well as a music source 4 my hot tub and my side deck (6 pairs total).
Using an ocelot and a secu 16 (supported by HAL2000, but not AHP as far as I know), I've wired contact closures on the sima so that Lexxie can control which speakers are on and which are off. The outdoor speakers only come on when I ask for them and they shut down if she plays a phone call announcement so that my neighbors don't hear who's calling me.
I will eventually upgrade my gate/mixer combo to a shure automixer which has logic control bus outs and will allow me to wire into the ocelot so that lexxie can tell which mic I just spoke thru and turn on only that room's speakers to answer.
Extraneous noise will b a problem until some significant improvements are made in SR technology. Imagine if someone taught you to recognize a few words of Farsi (not what they mean, just to recognize them) and then said "When you hear those words, raise your hand." If one person is speaking Farsi, slowly and clearly, you will very likely be able to pick out the words with pretty good acuracy. If 5 people are speaking Farsi at the same time, it becomes much more difficult. If 5 people are speaking Farsi at the same time and one of them is playing a recording of the Farsi equivalent of Britney Spears on their stereo... This is what we're asking speech recognition engines to do. Some improvement can be made by optimizing the input system (mic type, gating, eq settings) for the frequency range and volume of average human speech, but this only helps so much. The idea has been tossed around to encode music and television programs (or possibly the hardware that plays them) with a code that is outside the range of human hearing, that would cause a special microphone to ignore the sounds coming from this source, but that would be very expensive and take a long time to implement. A way around it (which I plan on setting up once I get my old house intercom system working) is to wire an interupt controller to the intercom button and use the intercom mic (overiding the open air setup) when the noise level in the room is too great. Some would say "then why not just use the light switch?" but the intecom system would still allow use of scenes and macros. Another option (as a backup to use in a noisy room) is the smarthome control maxi 4071. This is like the x10 maxicontroller on steroids and doesn't look like something you'd see on an episode of the rockford files. I'm trying to figure out how to wall mount one where the light switch used to be (using a lamp module for lights and an app module for ceiling fan - so the switch is just there to fill the whole) in my bedroom. I have one by the futon in the living room and will probably purchase another for the recording studio by the end of the year. My first attempt is always on voice and then I go to a fall back if voice doesn't work quickly enough.
An interesting note: Lexxie's performance seems to be inversely proportional to how impressive I need her to be at the moment. i.e. if I'm trying to show off for someone who's over to the house for the first time, she is less likely to work properly, but bread always falls buttered side down and your dog will only sit on command when no one else is around to see, it's just the way the universe works.
HAL has 4 levels of software, HALBasic, HALDeluxe (available in a kit at Fry's), HAL2000, and HALPro (available only thru contractors/installers). There are some problems, but I currently get about 85 - 90% acuracy, regardless of my location in the room. You have to pronounce things as if you don't speak english and are saying things phonetically (Example - pronounce Thank you as Tha-ngk Cue) to get a high level of accuracy on my system, but it's on an older computer with only an athlon 2300 and 1 gig of RAM. I'm lead to understand that a better chip significantly improves the speech recognition engine. She tends to confuse the words "on" and "off", so to get around this, I'm writing macros to respond to "Kill (unit)" as the eqivalent of Turn Off (unit). It's very timeconsuming to have to write this macro for each of 112 units, so I will be in touch with their customer support people (who are uber-responsive) to ask them to create a way to simply change that recognition phrase. The patrol their message boards hevily and always respond to suggestions from their users, but I digress. You can call HAL (Lexxie) from your cell and (with the voiceportal modem) use either dtmf or voice to control it. There is an internet interface (with included server) and even an app to allow control from my sidekick 2's browser directly. With HALi (a free add on similar to an SDK) ypu can write your own apps in perl (and I believe c++). Many user's have done so and made their apps available free of charge.
The cheap (think guitar center) nady and samson mics will reject any non protocol signals. If the signal is strong enough to overpower the matched transmitter's signal, you will experience a loss of audible signal or possibly static. The "other" signal, however, will not come thru as audio there for should not allow anyone else control of your automation. If your connection from the wireless receiver to the computer is not shielded/balanced, you may pick up audible rf from this, but the frequency range is such that you will most likely hear a tejano radio station regardless of the fact that there are no tejano radio stations within 1000 miles of your location. This is one of the great mysteries of the audio world.
Hope this helps and feel free to ask for anymore specifics you need.
PS. Sometime in the next couple of months (when my theatre is dark) I'm going to try all the mics in my rental inventory to see if any of them work beter than the pzm's. I expect I can get a slightly better result with the Audio Technica 853's that I use for hanging choir mics, but I'll post to this thread with my results...
Author
Topic: BXVC (Bill's X10 Voice Commander) (JScript version) (Read 345163 times)