CQC's upcoming 'AI' capabilities

Dean Roddey

Senior Member
So we've been working on a big new feature for CQC in the voice control area. We have Echo support and that's fine and it does what it does. But, we also wanted to have another option that met the criteria:
 
1. Easy to set up, with lots of 'out of the box' functionality
2. Purely local, no cloud
3. Much more in the 'AI' vein, not just a simple ask a question get a response type thing, so conversational in nature
 
We've had for some time a 'room configuration' option, where you can define your rooms and indicate what hardware is in each. We can use that to auto-generate a nice set of graphical touch screen interfaces for each configured room.
 
So this new AI is also based on that same configuration, and understands all of the functionality it provides (or it will very soon when we finish adding it all.) That means that we will be able provide this scenario:
 
1. Install CQC
2. Install the drivers for your devices
3. Auto-generate nice touch screens
4. Get nice AI style voice control with no extra work
 
CQC has something of a reputation for complexity, because it is primarily focused on flexibility for pro installations. But the above scenario will allow you to get a nice setup for core functionality without a lot of work.
 
Anyhoo, I just wanted to throw out this early demo of the new features, to get some feedback. If changes are to be made, then better earlier than later. This just demonstrates three different conversations that show how it works, with some text annotations to let you know what is going on.
 
http://www.charmedquark.com/Web2/PostImages/5_1Previews/JarvisPreview1.mp4
 
 
This is based on the Microsoft Kinect sensor, so it's all local. No security issues. The voice used for text to speech is just one of the standard ones MS provides, you can get better ones if you want.
 
Anyway, let me know what you think. We will of course keep our Echo support as well, so this is to provide another, very different option.
 
Great news Dean! 
 
I am an MS SAPI user and have collected voice fonts since the early 2000's.  Today I have domestic and international voice fonts.  I always liked having a choice.  Currently running SAPI in an Oracle VB on the Ubuntu mothership as Linux TTS appears to still be very ancillary.
 
The Windows 10 SAPI VR engine appears to be much faster and easier to train than the old stuff.  Tinkering with it here using the Kinect V1 device.  Did notice that new W10 release (or pre release) with an upgraded integrated Cortana appears to have put in some deep teeth to the MS OS as it is a tad slower now (to me).   Is your AI using the updated Cortana?
 
Pretty neat. Is there a way to stop it saying "will there be anything else?" Maybe it's needed, but just listening to the demo and I'm already annoyed. Haha.
 
Yeh, the discussion was that that would be an optional thing. For less technical users it's probably a good thing. For folks who use it more often, they should probably be able to turn it off. If you just speak another command within 10 seconds or so, it'll continue the conversation. If not, then it will ask if you want something else, to give you a chance to continue. Else it ends the conversation.
 
What I've ended up doing is:
 
1. You say a command or query
2. It will respond with success or failure
3. It will then wait for a bit for you to say another one
4. If so, goto #2
5. Else it will ask if you need anything else, it will wait for a shorter period of time for a new command
6. If so, goto #2
7. Else it will say one of the 'call me if you need me' variations and go back to waiting for the keyword

So it will only prompt you if you don't provide a new command or query in a normal sort of conversational period of time, just to give you a chance to continue the conversation.

You can also cancel the conversation at any time by one of the 'that will be all' type phrases.
 
After watching the demo a few more times I am increasingly impressed with how natural the interaction feels. What an excellent piece of work!
 
Thanks! I'll put up a new one here in a couple days, that takes it to the next step, probably pretty close to what it'll actually be upon initial delivery.
 
Hi,
 
Looking good.
 
Are the phrases spoken by the system changeable?
 
I have several Echo devices and I can chose to have a simple "beep" sound to indicate that "Alexa" has been recognized and is waiting for a command or has finished the previous command.
 
I personally would get very tired of listening to phrases, which take a bit of time, when a simple beep or two would suffice.
 
Just my two cents.
 
Frederick
 
It's been changed now. I just waits for a little bit to give another command, once a conversation has started. If you don't, it'll ask then if you want anything else, then wait a shorter bit, then excuse itself if not. Otherwise you can just continue giving commands or queries. Or you can dismiss it explicitly at any time as well.
 
That should work pretty well.
 
Back
Top