Amazon Echo to HA Controllers

jkmonroe said:
The GEM?  What package/CTs and interface?  It is next on my list, too.
 
how about I just ship it to you, you figure it out, and if you like it you pay me. If not you return it. Its been in the same box it shipped in for 2 years now. Soccer is just so much more fun than running my 200th wire in this house.
 
ChrisCicc said:
To be fair, I described the normal app voice path. You've hacked something together that cannot be made a commercial production product. It's another apples to oranges comparison. All the voice commands you have to manually enter as well. CastleOS does it all automatically just be seeing the groups and devices you have in the system.

Most people have lock screens. Most people don't use tasker. They use apps. Wink app. Vera app. HomeKit app. As you've found, you need a trigger word. So now you've built an Amazon Echo equivalent, not a CastleOS equivalent. Not to mention the lag inherent in the cloud... 
uh, huh?  My "hacking" consisted of:
1) Buy a device and put it on my nightstand. (Nexus 7)
2) Install an app. (Tasker)
3) Install another app. (AutoVoice)
4) Set up a profile & event per command.
5) Type in what words I use. (technically I didn't even do this, I said it out loud, AutoVoice transcribed it)
6) Type in what I want to happen. (ie, the CQC URL to hit. Then inside CQC, the list of actions I want)
 
Thats it. I didn't even have to do any programming.
 
CastleOS might eliminate step 4, but it can't magically know what I want to do based on what I say, I have to manually enter *something*.  I rarely just say a single command at a time, that would be silly. IE, I wouldn't say "lights off. stereo off. turn down the thermostat."  I'd say "Okay Google, I'm leaving you".    AutoVoice then sends CQC the shutdown command, CQC knows what I want to do. Similarly, I'd have to tell CastleOS what I want to do. There's no way around that, because what I want to do upon leaving the house (in a semi-urban location in northern california) is far different than what a texan or a new yorker would want to do, by virtue of having different houses, different devices, different climates.
 
I also don't see why I care about Echo vs CastleOS equivalency. Then again, i'm not like most HA folks. I don't do HA because I inherently enjoy this, I do it to achieve an outcome. When I say "I'm leaving", I don't care if its an echo speaker, kinect, or trained monkeys going around and shutting the house down, I just want the house to shut down. What I want is reliability and not have to repeat myself. 
 
re:Lock screens, why would anyone need a lock screen on a fixed location device like an inhouse tablet? You could easily disable that, matter of fact you *should* disable that. You don't have a lock screen on your kinect, do you?  The only reason to put a lock screen on is if you want to use it outside the house, and if you want to compare apples:apples, the only thing to install on the android device is tasker & autovoice. 
 
BTW, I haven't noticed a cloud lag when I'm in a fixed location with broadband uplink like my house. I only ever get it while driving.
 
I'm not trying to be disagreeable, but I still don't see what CastleOS offers beyond a supposedly better VR mechanism, but one reliant on hardware from a single manufacturer (M$ Kinect). In my case, I have no cloud latency but awful acoustics on the Kinect voice, so relying on that is actually a step backwards for me.
 
IVB said:
how about I just ship it to you, you figure it out, and if you like it you pay me. If not you return it. Its been in the same box it shipped in for 2 years now. Soccer is just so much more fun than running my 200th wire in this house.
 
uh, huh?  My "hacking" consisted of:
1) Buy a device and put it on my nightstand. (Nexus 7)
2) Install an app. (Tasker)
3) Install another app. (AutoVoice)
4) Set up a profile & event per command.
5) Type in what words I use. (technically I didn't even do this, I said it out loud, AutoVoice transcribed it)
6) Type in what I want to happen. (ie, the CQC URL to hit. Then inside CQC, the list of actions I want)
 
Thats it. I didn't even have to do any programming.
 
CastleOS might eliminate step 4, but it can't magically know what I want to do based on what I say, I have to manually enter *something*.  I rarely just say a single command at a time, that would be silly. IE, I wouldn't say "lights off. stereo off. turn down the thermostat."  I'd say "Okay Google, I'm leaving you".    AutoVoice then sends CQC the shutdown command, CQC knows what I want to do. Similarly, I'd have to tell CastleOS what I want to do. There's no way around that, because what I want to do upon leaving the house (in a semi-urban location in northern california) is far different than what a texan or a new yorker would want to do, by virtue of having different houses, different devices, different climates.
 
I also don't see why I care about Echo vs CastleOS equivalency. Then again, i'm not like most HA folks. I don't do HA because I inherently enjoy this, I do it to achieve an outcome. When I say "I'm leaving", I don't care if its an echo speaker, kinect, or trained monkeys going around and shutting the house down, I just want the house to shut down. What I want is reliability and not have to repeat myself. 
 
re:Lock screens, why would anyone need a lock screen on a fixed location device like an inhouse tablet? You could easily disable that, matter of fact you *should* disable that. You don't have a lock screen on your kinect, do you?  The only reason to put a lock screen on is if you want to use it outside the house, and if you want to compare apples:apples, the only thing to install on the android device is tasker & autovoice. 
 
BTW, I haven't noticed a cloud lag when I'm in a fixed location with broadband uplink like my house. I only ever get it while driving.
 
I'm not trying to be disagreeable, but I still don't see what CastleOS offers beyond a supposedly better VR mechanism, but one reliant on hardware from a single manufacturer (M$ Kinect). In my case, I have no cloud latency but awful acoustics on the Kinect voice, so relying on that is actually a step backwards for me.
 
You're describing both an atypical and a narrow use case. No doubt your setup works for you, but it doesn't mean that will translate to others. To you, steps 4, 5, and 6 aren't difficult. You've programmed in what you need and it's good to go. But most consumers cannot do that. Most consumers do not want to do that. In addition, depending on the installation, CastleOS can accept well over 100,000 commands. It's all automatically built, including your shut down commands, based on groups, devices, scenes, and events. No programming needed of any kind.

This is a DIY site, and of course DIY solutions are going to be preferred by many. But that doesn't mean they are the best or simplest way of doing it :) 
 
ChrisCicc said:
You're describing both an atypical and a narrow use case. No doubt your setup works for you, but it doesn't mean that will translate to others. To you, steps 4, 5, and 6 aren't difficult. You've programmed in what you need and it's good to go. But most consumers cannot do that. Most consumers do not want to do that. In addition, depending on the installation, CastleOS can accept well over 100,000 commands. It's all automatically built, including your shut down commands, based on groups, devices, scenes, and events. No programming needed of any kind.

This is a DIY site, and of course DIY solutions are going to be preferred by many. But that doesn't mean they are the best or simplest way of doing it :)
 
okay now its my turn to not be clear; Even with 100K commands, if I say "shut the house down i'm leaving", how can CastleOS know what I mean vs what JKMonroe means?  CQC doesn't require any programming, but you do need to list out the devices you want to shut down. Same way - you need to tell CastleOS what list of devices you want to shut down. And, if you can't handle step 4, which is literally A)say command out loud, B)type in a URL, then you have no business using a computer much less creating an HA system.
 
BTW, the more I think about it, the more dangerous it is to rely on a single hardware device by a single manufacturer. I first started doing my HA system 10 years ago, the # of major manufacturers who started and stopped doing HA, or locked down their products, is high.  If M$ decides to shut down using Kinect in this way (or worse, creates a whole new product line and deprecates this), what would the impact be?
 
For me, i'm protected against planned obsolescence as VR is abstracted out to being just a triggering mechanism. I'm sick of companies shutting down or going a different way. *cough cough* Intermatic *cough cough*
 
IVB said:
okay now its my turn to not be clear; Even with 100K commands, if I say "shut the house down i'm leaving", how can CastleOS know what I mean vs what JKMonroe means?  CQC doesn't require any programming, but you do need to list out the devices you want to shut down. Same way - you need to tell CastleOS what list of devices you want to shut down. And, if you can't handle step 4, which is literally A)say command out loud, B)type in a URL, then you have no business using a computer much less creating an HA system.
 
Voice recognition in home automation is only an extension of what is currently done in an app. For instance, you may have a "shutdown" house scene or event that you can trigger through a regular app interface. To build that scene, you've listed all the devices you want turned off, dimmed, on, etc, when the scene is run. Why should you need to relist all that for a voice command? You don't need to with CaslteOS.
 
In my own home, I have an "All Lights" scene. If I press off on the scene in the app, all the lights listed go off. If I say out loud as I'm walking out the door, "house all lights off" or "house turn off all lights", it runs just as it would if I pressed "off" in the app. The only thing I had to do to get that voice command to work, was install the CastleOS app, plug in a Kinect, and log in for the first time :)
 
And about your comments with step 4, that's fine for you, but not the millions of new home automation consumers joining the club. If we took that attitude as a company, we'd be laughed at.
 

IVB said:
BTW, the more I think about it, the more dangerous it is to rely on a single hardware device by a single manufacturer. I first started doing my HA system 10 years ago, the # of major manufacturers who started and stopped doing HA, or locked down their products, is high.  If M$ decides to shut down using Kinect in this way (or worse, creates a whole new product line and deprecates this), what would the impact be?
 
For me, i'm protected against planned obsolescence as VR is abstracted out to being just a triggering mechanism. I'm sick of companies shutting down or going a different way. *cough cough* Intermatic *cough cough*
 
In some cases that's true, but not only is the Kinect not going anywhere, there are already tens of millions of them shipped. There have literally been more Kinects shipped that Insteon or Nest products. They are available on Craigslist and Ebay for $15. Getting them isn't an issue. But more importantly, the same foundation we've built is being ported to Cortana and Xbox. Not to mention upcoming Android and Android Wear voice support. The Kinect is only one way to interface via voice, and the best at doing "out loud" voice control. If MS replaces Kinect with something else, they'll do it via the same software they currently do. The Kinect is just a microphone array, in the end. The real meat of it is in the software...
 
Edit: Now that I think about it, relying on Google or Amazon clouds for voice recognition brings far greater risk of being locked out or shut down than our method. We don't need the cloud. Once shipped, our software can run forever. So the risk is much lower, in actuality, than your solution. 
 
IVB actually has it correct.  The apps are called autovoice/tasker and they are commercial apps.  You configure them and they really do work that way, no hacking required.  It's not like the Siri solution which requires hacking, intercepting certificates, etc.  And it is nice with the MotoX chip, what you get there is you don't have autovoice eating your battery life to deliver always on listening.   The drawback is you have to be a few inches away, no far field microphone, which is what Echo/Kinect apparently bring to the equation. The benefit to android is you can take it with you, which you can't do with the Kinect/Echo.
 
The biggest drawback of Google Now, Echo, Siri, Cortana that I see are that they rely on the cloud so introduce latency, which can be noticeable and also the privacy concerns.   A local solution like Castle would be better in that regard but you lose any AI so are stuck with fixed grammars.  It will be interesting to see how Win10 changes that, if at all.   I am hoping we will see more apps with VR, MS really is ahead of the game there.  They have been doing VR research since before Google existed and that is baked right into the OS.
 
wuench said:
IVB actually has it correct.  The apps are called autovoice/tasker and they are commercial apps.
 
No automation company has shipped this combined solution for a reason. I understand those are apps you can download, but what I'm referring to is a company like CastleOS using them to enable voice functionality. That isn't feasible. I agree with the rest of what you said though. Ultimately, we'll support both types of voice control, depending on what the platform supports. More choice for our users is always a good thing :)
 
Ok, i'll redact my comments too.. ;)  I do like the local solution best.  I think the next step is for it to be affordable over the whole home.  We can't be putting PC's all over with a big ass mic, and centralized systems are tricky with gating, etc.   Maybe Win10 will bring local speech to a small enough form factor to be feasible.  And if they could just take Kinect and throw away the stuff we don't need like cameras, motors, etc.  That would be ideal.  
 
wuench said:
Chris, this is where you get yourself in trouble talking about stuff outside of your experience.   This is what people have a problem with around here.
 
How is it outside my experience? Am I not an expert in the voice interface field? Has anyone else brought to market a whole home voice control system? 
 
The millions of consumers entering the market are looking for what I look "ehrmergerd lookie what I can do with ma voice! Ain't that cool! Alexa should I fake my orgasms?" (Although I own a ludicrous amount of AAPL stock so I encourage this behavior)
 
I wrote up a guide years back on the overall effort of putting in a home automation system. Its not custom programming, its everything all-in.  I'm not talking about some cutesey little thing where you can control a few lights, a nest, and maybe oooh - your sonos. (Which I also own btw but it feeds my nuvo concerto whole house audio system).
 
What I mean is home automation, and controlling most of the devices in your house. And, having your system monitor those devices and take action.  Stuff like, the security system was armed in away mode but not everything was turned off, the system should proactively do that as I probably just forgot to say "i'm leaving". (or tap the NFC chip with my phone).
 
For someone doing actual home automation where their house does stuff without their intervention, or with their voice, typing in a URL isn't actually difficult.  And it represents far less than 1% of the overall time investment to construct it. 
 
Millions of people are NOT doing that. They don't want to do that. They don't want to do home automation themselves, can you imagine millions of people even attempting to install an HVAC thermo? 99.9999% of people will just call their HVAC guy to do it, despite it being 5 wires color coded, and maybe a cat5e if they hardwire it like I did.   Or adding another motion sensor to their security system so that they can use both motion and voice to determine what to do and where. Thats even less, thats only 4 wires. 
 
The millions of consumers entering the market are looking for what I look "ehrmergerd lookie what I can do with ma voice! Ain't that cool! Alexa should I fake my orgasms?" (Although I own a ludicrous amount of AAPL stock so I encourage this behavior)
 
I wrote up a guide years back on the overall effort of putting in a home automation system. Its not custom programming, its everything all-in.  I'm not talking about some cutesey little thing where you can control a few lights, a nest, and maybe oooh - your sonos. (Which I also own btw but it feeds my nuvo concerto whole house audio system).
 
What I mean is home automation, and controlling most of the devices in your house. And, having your system monitor those devices and take action.  Stuff like, the security system was armed in away mode but not everything was turned off, the system should proactively do that as I probably just forgot to say "i'm leaving". (or tap the NFC chip with my phone).
 
For someone doing actual home automation where their house does stuff without their intervention, or with their voice, typing in a URL isn't actually difficult.  And it represents far less than 1% of the overall time investment to construct it. 
 
Millions of people are NOT doing that. They don't want to do that. They don't want to do home automation themselves, can you imagine millions of people even attempting to install an HVAC thermo? 99.9999% of people will just call their HVAC guy to do it, despite it being 5 wires color coded, and maybe a cat5e if they hardwire it like I did.   Or adding another motion sensor to their security system so that they can use both motion and voice to determine what to do and where. Thats even less, thats only 4 wires. 
 
wuench said:
 
Android is not a whole home voice control system. Android devices have never even shipped with a far field microphone, never mind a far field microphone array, as far as I know.
 
(Edit: even then, all it does is convert speech to text. It has no home automation capabilities.)
 
wuench said:
Ok, i'll redact my comments too.. ;)  I do like the local solution best.  I think the next step is for it to be affordable over the whole home.  We can't be putting PC's all over with a big ass mic, and centralized systems are tricky with gating, etc.   Maybe Win10 will bring local speech to a small enough form factor to be feasible.  And if they could just take Kinect and throw away the stuff we don't need like cameras, motors, etc.  That would be ideal.  
 
Everything is getting smaller. Two years ago, small form factor PCs were the size of Shuttle PCs. Then the NUC came out, then NUC like competitors went solid state, and got smaller still. Then Intel announced the Compute Stick. Like all tech, it's shrinking rapidly.

Regarding shrinking the microphone array, I agree with that. We have something in the works regarding that, but it will be many months before I can begin to say more :)
 
No, what I am saying is you obviously don't have any experience with Android's voice solution.  That's all.  You got a great product in Castle, I have played with it and am looking forward to see how it matures.
 
wuench said:
No, what I am saying is you obviously don't have any experience with Android's voice solution.  That's all.  You got a great product in Castle, I have played with it and am looking forward to see how it matures.
 
I'm not referring to those apps ability to work well (or not, as the case may be). I'm referring to our ability to ship integrated support for them, with zero configuration for the end consumer. It isn't possible, that's fact. 

Hobbyists can use them with the CastleOS API all day long though... 
 
Thank you for the kind words about CastleOS too!
 
Back
Top