Whole House Audio - Voice Control

[DanielHeth]

I'm putting my HomeSeer onto a more power system in order to install multiple sound cards.  My goal is to have a computer with speakers in every room... then homeseer could be controled to send audio to each room individually or all together.
 
I can handle the software side of things... either finding something or programming something myself.
 
What i'm looking for help with is understanding the hardware needs.  Obviously i'll need speakers and microphones in each room... but do i just wire the speakers directly up to the sound cards?  Do i need to add any sort of converters, or surge protectors on the audio lines, etc...
 
Does anybody have experience doing this?

 

[elvisimprsntr]

If you really want a computer in every room, may I suggest that you look at an inexpensive Raspberry PI ARM computer which has both audio and video output. There are people who have them set up as an audio streaming device which you can stream from any location. While the RPI does not have a built in microphone you can add a USB one. Speech recognition is difficult enough as it is. You really need close proximity to the user and noise canceling microphones. This is why I use my mobile devices for speech recognition.

http://youtu.be/PXmCiaRc9XU

 

[electron]

Voice Recognition can be pretty tricky.  I even purchased a tracking microphone array, but ended up returning it because it just wasn't good enough.  Some people have it working by installing really sensitive microphones in each room, which might do the job (I actually purchased this one as well, and it is extremely sensitive for sure).  The other problem is background noise.  Once you have other people talking in the background, or watching TV, it reduces the usability significantly.
 
I really like the siriProxy approach, and if I ever try VR again, it will probably be based on that type of system.

 

[BraveSirRobbin]

Are you just going to use amplified speakers and have them sitting on a cabinet or desktop, or are you planning on remote amps/speaker mounting?

 

[DanielHeth]

elvisimprsntr: nice job on the Siri proxy approach.  Do you have any write ups i can learn more about how you set that up?
I'm in the early research phase of adding VR to my system and looking for various options.
I'm assuming your not required to be on Wifi for everything to work.
 
The microphones are going to be tricky if i continue towards the whole house approach.  I was also considering small lapel mikes but then i'd have to always remember to clip something on and change batteries. 
 
BraveSirRobbin:  I'm looking at putting in flush mount speakers like in standard offices.  I like the clean non intrusive approach and think those would be the simplest "look" to add to rooms.  As for wiring them up, this is where i have no idea and asking.  You do have a point about amplification.  Is there some sort of in-line amp i can add to each speaker so the seperate signals from each sound card output is increased to appropriate levels?

 

[elvisimprsntr]

elvisimprsntr: nice job on the Siri proxy approach.  Do you have any write ups i can learn more about how you set that up?
I'm in the early research phase of adding VR to my system and looking for various options.
I'm assuming your not required to be on Wifi for everything to work.

 
All the details are in the video description of you open up the video on YT.   Video also has links to my SF hosted RPi SD card image with SP pre-installed, and my SP plugins on my GH page.    I just completed a re-write of my voice controlled IP2IR control plugin to make it auto discovery. https://github.com/elvisimprsntr/siriproxy-redeye    
 
I can use SiriProxy to control my home remotely thru a VPN connection which is easy if you know what you are doing.    I can say "we're on our way home" from a 1/4 mile away and Siri will disarm the alarm, turn on the lights, and open the garage door.  

 

[ChrisCicc]

Hi Daniel, please take a look at our software, CastleOS. Depending on your needs it may not be able to do what you need as HomeSeer supports more devices and protocols, but our voice control is unparalleled. Just launched in January. www.CastleOS.com

 

[BraveSirRobbin]

You can't run the sound directly out of the soundcards (pre-amp output) to the speakers without some sort of amplification.  I guess I'm starting to wonder the merit of using multiple sound cards vs getting a true self contained distribution system.

 

[DanielHeth]

Thanks for the additional links... i'll look further into that elvisimprsntr.
 
ChrisCicc:  The site looks great and the software is interesting, but my main problem with it is there is no mention of the physical components required to do the things your software can control.  HomeSeer has a good mix of their software with a huge list of compatible hardware that they sell and have demos on how to setup.  My marketing to me, you're approaching a DIY'r directly which requires more than a flashy site.  I like the site, don't get me wrong... but as a DIY'r i need to know a lot more about how the rubber meets the road with your solution.  How are you going to convince me to abandon the $500 i already spent on homeseer, a long standing proven tech for something unknown.  I'd also recommend a long trial version.  Remember you're marketing to people who already have systems.  Some of us would devote resources to your solution, which includes moving some HA hardware over to it, but that takes time.  I'd recommend 60 or even 120 day trials... 2 weeks isn't nearly enough to be taken seriously by me, sorry. 
I will keep you in my favorites list and revisit your site to see how you grow, but for now i'm along way away from switching.
 
BraveSirRobbin:  This is what I'm looking for as far as guts... the details like this are very helpful.  I obviously remembered that non-powered speakers don't work very well from sound card outputs... but sometimes even the best of us have momentary lapses. 
What i'm after with regards to the multiple sound cards is the output control.  The ability to write a program and have the sound sent down a specific card (ie: to a specific room).  The built in sound card can seperate the front speaker output from the rear speaker output, giving me two rooms.  Many newer sound cards can do the same.  If anybody has seen a card with more than two outputs controlled separately... that would help too.   Obvoiusly it still comes down to hardware... what recommendations do you guys have for powered speakers? 
The easiest to install would be flush mount speakers with standard speaker wire leading back to the wiring closet and into the system.  It sounds like i'll need to add an amp... but it'd be nice if it was a cheap simple device... which the speaker gets plugged into and into the sound card along with power.  I'll have to do some hard looking to try and find this device.

 

[ChrisCicc]

Daniel - agreed on the marketing. We've been working on a more detailed info page we'll have up soon. The short version is right now it works with Insteon, and will work with Z-Wave in about a month. All it takes is a Windows PC, same as HomeSeer. As far as the trial, the March 1st date is the end of the first beta period, not the end of a trial. It will continue to be free while in beta, and when out of beta you can take advantage of both the trial period on the commercial release, and the 75% discount we are offering to all beta users ($49.99 total). 
 
As far as the multi-room audio, we're working on functionality as part of our Kinect package that includes output via sound cards as well as distributed systems like Russound and Sonos. Without a Kinect, it's a multi-room music player. With a Kinect - well I can't say in detail yet but it's a lot more Your idea for sound cards would work, and in fact, you can get ones with four and sometimes five or six outputs - they allow you set all analog outputs as individual line outs rather than FL FR RL RR C etc. Check out the high-end Creative cards. 

You might also want to consider buying Sonos WiFi speakers, as they are UPnP based, any software can output audio through them pretty easily. Sonos has also stated they do not intend to block this functionality. This may cost a little more, but is very elegant.

 

[BraveSirRobbin]

Take a look at JohnW's blog and what he did with now-playing.  I believe he uses multiple sound cards (the Delta?) that you can get on EBay.
 
As far as software, I agree, just stick with Homeseer as you suggested.

 

[aksor]

I have a central server that runs Homeseer and Jriver MC with two soundblaster 5.1 audio cards each and I'm running the Kx driver to turn the sound cards into 6 individual stereo pairs,
 
In JRiver MC I set up multiple zones and assigned them to the sound card outputs.
 
I also run 6 instances of homeseer speaker's and assign it to each of
the jrivermc zones this way i can have both multizone music and voice
announcements.
 
The output of the sound cards go to a multi channel amp then to speaker pairs in the rooms.
 
In homeseer you can create events to make announcements to all or specific speaker instances, or you can even write scripts to control where to make the announcement.
 
if you need more info feel free to contact me
 
works great

 

[AutomatedOutlet]

Voice is very tricky like BSR said. If you insist on doing it, you will most likely need an audio mixer like the Crown 4 or 8 channel mixer.
 
If you want to talk to someone that has expensive experience with this, please call Bill in our office. He used to work for the company that invented this technology - Home Automated Living. 
 
He can be reached in our main office - 214-245-4594.

 

[naething]

I have a very cheap way of addressing your problem (at least the distributed sound / TTS part). I am quite fond of Logitech Squeezeboxes. They have been discontinued, but the code is open source so you can make your own.
 
I have custom python code to pause my squeezeboxes, speak a command with TTS, and then resume playback. But as you have Homeseer I believe the plugin will do that natively. The best part about this approach is you can test it out right now - just download SqueezePlay on a few computers and set up the Homeseer plugin.
 
This does not address the voice commands. I use my Android phone / tablets. That works well as will Siri. I doubt you will be happy with distributed microphones. It is very hard to deal with background noise.
 
Two cheap options for DIY Squeezeboxes (both ARM devices /w Archlinux Arm):
 
1) Raspberry Pi ($35) + SqueezePlay (open source, free)
2) Pogoplug E02 ($15) + Zalman ZM-RSSC Sound Card ($3.99, geeks.com) + 8GB usb flash ($5) + Squeezeplay (open source, free)
 
I have eight of the second option. The Raspberry will be easier to set up. If you go with #2 you will have to get a working sound config file which is a bit tricky, I am happy to share mine. You want to use the SqueezePlay player rather than SqueezeSlave. The later is restricted to 44.1kHz which causes problems with some sound cards.
 
This just gives you line level audio (as does your computer sound card approach). You then need to plug in some amplified speakers or an audio amp.
 
One option:
1. LP-2020A+ Lepai T-amp ($20 from Amazon)
2. Monoprice in-ceiling / in-wall speakers ($50 a pair).
 
I think that would give you pretty good bang for the buck. I would run the speaker wires from the ceiling / wall down to a keystone outlet in the room. Then hook up the amp from there.