Jump to content

Interface with Amazon Echo?


awzulich

Recommended Posts

Michel,

 

From what I read in the Amazon documentatios, literals should have only a few samples while custom slots show have as many possibilities as is possible/practical. if you allow 1, 2, 3, and 4 word "device names" or in my case "favorite names", I found that 1 sample for each case works best.  "Kitchen", "Back Porch", "Front living Room" "Rear porch high hats"...

 

"With literals, there's another drawback: what Echo responds with has to be altered." - Not sure what you mean by this. In my case I take what I am passed by the Echo and use a LCS function to find that best match.  If the data is even close to the favorite name it works well. If the data is totally off, then it either  returns with a random favorite or comes back with no match.

Link to comment

Michel,

 

From what I read in the Amazon documentatios, literals should have only a few samples while custom slots show have as many possibilities as is possible/practical. if you allow 1, 2, 3, and 4 word "device names" or in my case "favorite names", I found that 1 sample for each case works best.  "Kitchen", "Back Porch", "Front living Room" "Rear porch high hats"...

 

"With literals, there's another drawback: what Echo responds with has to be altered." - Not sure what you mean by this. In my case I take what I am passed by the Echo and use a LCS function to find that best match.  If the data is even close to the favorite name it works well. If the data is totally off, then it either  returns with a random favorite or comes back with no match.

 

Actually, testing shows that custom slots work well when the skill is not too large. It behaves just like a literal, with the added benefit that it helps recognition if what is spoken is close to one of the samples.

 

The problem we are facing are these:

1 - As we add more utterances, alexa starts having difficulty passing values to the custom slot. With only the turnOn turnOff utterances, it works very well. Just about any word spoken and correctly understood as seen on the echo app are passed in the custom slot to the skill. However, if we gradually add more unrelated utterances to the skill, alexa start dropping the last word like; light, lights, lamp, etc.. As we add even more, it stops passing simple words to the skill like "bard", "master, etc... Except for the words in the samples. I made sure that this is not due to words like "light" in the added utterances. I thoroughly tested with different blocks of utterances.

 

2 - As we add more utterances, we can save the skill (at least it appears as such), but we reach a point where underneath, something looks corrupted. Let's say I add many utterances, I save, everything looks good. It is there. But then, if we delete some of those utterances and save... they still work. I took such a skill, trimmed it to the point where I had only 1 utterance with 1 intent, and the skill still worked as if was never modified. So something is not updated underneath, but that only happens if the skill reach a certain size.

 

Both problems, which may very well be related, have been reported to Amazon.

 

The only possible short-term fix would be to add samples to the device list. That I know works.

 

Benoit.

Link to comment

AFAIK, Many if not most ISY users, use mostly Turn On, Turn Off, and Dim to control things. A native Alexa skill in a connected home (what Amazon calls a skill adapter) handles those commands very nicely for devices such as lights and switches. I have friendly names of 1, 2, and 3 words which Alexa seems to handle quite will to send rest commands for devices, scenes and programs. Using the skill adapter I do not have to say Ask Izzy or Tell Izzy.

 

Perhaps a reasonable solution is to use the Connected home skill adapter concept for those things it seems to handle quite well, and a formal skill (Izzy) for those things where different commands (other than Turn on, Turn off, dim) are desired. The skill (Izzy) could then be used to handle climate devices (thermostats); programs with commands of Run, Run If, and Run else; and other commands such as Open, Close. That should reduce the Utterance set for the skill.  

 

Just a thought 

Link to comment

I have gone into the portal to set my spoken names, but I notice that I am missing several of my scenes. Even after a refresh they are missing, also what about programs.

 

Ron

 

Hello Ronbo,

 

The refresh button only refreshes the table with content hosted on portal. This does not do the same as  "refresh device" did. It is now a somewhat a manual process where you decide which devices you want to expose to the skill (and eventually connected home), and of course the corresponding spoken.

 

Currently, programs work the same as before. They work with the name that you have on your ISY.

 

However, you can expect this to change. You will soon see a button added there to allow to define a spoken for programs.

 

Benoit.

Link to comment

Benoit,

 

Will we be able to tell the portal to exclude a device/scene/program from the skill?

 

I have over 200, and I only need less than 10 for my Echo (given its location in my house). I'm sure that by filtering devices/scenes/programs, a better, faster match to verbal requests can be made.

 

Thanks!

 

Michael.

Link to comment

Benoit,

 

Will we be able to tell the portal to exclude a device/scene/program from the skill?

 

I have over 200, and I only need less than 10 for my Echo (given its location in my house). I'm sure that by filtering devices/scenes/programs, a better, faster match to verbal requests can be made.

 

Thanks!

 

Michael.

 

Hi Michael,

 

The mapping of spoken to devices is now managed on portal. There is an "Amazon Echo" tool that has been added on the ISY page.

 

The skill can only use devices defined there. You can delete those that you don't need.

 

Benoit.

Link to comment

 

 

Hi Michael,

 

The mapping of spoken to devices is now managed on portal. There is an "Amazon Echo" tool that has been added on the ISY page.

 

The skill can only use devices defined there. You can delete those that you don't need.

 

Benoit.

When I tried to clear a spoken name, it gave me a 400 error. It also filled all spoken name fields in with the device name, most of which are structured for ease of use in the Admin Console, and are not suited for voice recognition.

 

As an aside... I think an ISY level option to have Alexa ignore the natural device names would be useful. I also think that spoken name fields should default blank.

 

That aside, are you saying I should delete the device on Portal?? Or just clear the spoken name field?

 

When I adjust the spoken name field on Portal - does this sync back to ISY? They don't appear to - which makes me wonder why the recent changes to allow them, and have them show up in the topology report....

 

Michael.

Link to comment

Hello Ronbo,

 

The refresh button only refreshes the table with content hosted on portal. This does not do the same as  "refresh device" did. It is now a somewhat a manual process where you decide which devices you want to expose to the skill (and eventually connected home), and of course the corresponding spoken.

 

Currently, programs work the same as before. They work with the name that you have on your ISY.

 

However, you can expect this to change. You will soon see a button added there to allow to define a spoken for programs.

 

Benoit.

 

Benoit,

 

I think I understand, but when I click on the scene tab to add a spoken word to a scene, not all of my scenes are listed.

Am I missing something?

 

Ron

Link to comment

When I tried to clear a spoken name, it gave me a 400 error. It also filled all spoken name fields in with the device name, most of which are structured for ease of use in the Admin Console, and are not suited for voice recognition.

 

As an aside... I think an ISY level option to have Alexa ignore the natural device names would be useful. I also think that spoken name fields should default blank.

 

That aside, are you saying I should delete the device on Portal?? Or just clear the spoken name field?

 

When I adjust the spoken name field on Portal - does this sync back to ISY? They don't appear to - which makes me wonder why the recent changes to allow them, and have them show up in the topology report....

 

Michael.

 

 

Hi Michael,

 

I was suggesting to delete the devices on portal. That's how you opt out a device from the Echo.

 

You are right, they are not synched with the ISY. It's now totally separate.

 

We could have started with an empty table instead, but we figured it would be best to start with those from the ISY so that the move to production is not disruptive.

 

Benoit.

Link to comment

Benoit,

 

I think I understand, but when I click on the scene tab to add a spoken word to a scene, not all of my scenes are listed.

Am I missing something?

 

Ron

 

Ronbo,

 

They should all be listed, unless you type something in the field, which has the effect of filtering on the list shown.

 

Are you sure you are missing scenes?

 

Any one else seeing this behavior?

 

Benoit

Link to comment

Ronbo,

 

They should all be listed, unless you type something in the field, which has the effect of filtering on the list shown.

 

Are you sure you are missing scenes?

 

Any one else seeing this behavior?

 

Benoit

 

Benoit,

 

Definitely missing several scenes, rebooted ISY hit referesh, still missing scenes.

 

Ron

Link to comment

Benoit,

 

Definitely missing several scenes, rebooted ISY hit referesh, still missing scenes.

 

Ron

 

Hi Rombo,

 

Please click on Add scene. There, you will be able to add scenes to your list.

 

The refresh button does not synchronize with the ISY. It just refreshes the UI with the content on the portal server.

 

Benoit.

Link to comment

Hi Michael,

 

I was suggesting to delete the devices on portal. That's how you opt out a device from the Echo.

 

You are right, they are not synched with the ISY. It's now totally separate.

 

We could have started with an empty table instead, but we figured it would be best to start with those from the ISY so that the move to production is not disruptive.

 

Benoit.

Thanks Benoit,

 

I just found the (tiny) red 'x' to remove a device from the Alexa mapping table... - and noticed the 'Delete All'.

 

So, I deleted all then added the scenes I want controllable via Alexa. I'll test when I get home this evening.  It does make more sense managing this directly on the portal vs on the ISY. A question remains though - what is the purpose of the 'spoken' field on the ISY now? Is this for the connected home Alexa integration whenever Amazon approve it - or some other purpose?

 

One (minor) comment...

 

There is an 'Add Device' button. Why is 'Add Scene' labeled simply 'Scene'?  Shouldn't it be 'Add Scene'?

 

Michael.

Link to comment

Thanks Benoit,

 

I just found the (tiny) red 'x' to remove a device from the Alexa mapping table... - and noticed the 'Delete All'.

 

So, I deleted all then added the scenes I want controllable via Alexa. I'll test when I get home this evening.  It does make more sense managing this directly on the portal vs on the ISY. A question remains though - what is the purpose of the 'spoken' field on the ISY now? Is this for the connected home Alexa integration whenever Amazon approve it - or some other purpose?

 

One (minor) comment...

 

There is an 'Add Device' button. Why is 'Add Scene' labeled simply 'Scene'?  Shouldn't it be 'Add Scene'?

 

Michael.

 

Hi Michael,

 

Connecte home will use this same table. 

 

Now, the spoken field on ISY has no purpose related to portal. It can perhaps be used for other home automation needs, if integrated with something that speaks or can accept spoken requests, or just be ignored.

 

About the "Spoken" not having Add... well, the idea is that there will be Programs as well. It may not be too obvious, but those two buttons form a buttonset. For esthetic & real estate reason, I opted for a button set where you have 3 places to click on to Add devices/scenes/programs. When you see "program" appear there, having "Add" in front of all 3 will look redundant.

 

Benoit.

Link to comment

Hi Rombo,

 

Please click on Add scene. There, you will be able to add scenes to your list.

 

The refresh button does not synchronize with the ISY. It just refreshes the UI with the content on the portal server.

 

Benoit.

 

 

 

Benoit,

 

if I click the add scene and manually type in the scene name it says device not allowed... I would expect to see the scene in the drop down list. and I only see some of them

 

Ron

Link to comment

Benoit,

 

if I click the add scene and manually type in the scene name it says device not allowed... I would expect to see the scene in the drop down list. and I only see some of them

 

Ron

 

Hi Ron,

 

When you start typing, it should bring you a dropdown with scenes that match what you type. Or alternatively, if you the field is empty and you click on the arrow, you should see all your scenes in the dropdown.

 

If that's now what is happening, I'm starting to think that it may be a browser specific issue. What browser are you using?

 

Benoit

Link to comment

Benoit,

 

I have not yet looked at the Portal.  When I have a UI with a "Button Set"  I generally put a frame around the button set and then put a title in the frame like "Add" which then pertains to the entire set in the frame. Users seem to like that approach.

Link to comment

Benoit,

 

I have not yet looked at the Portal.  When I have a UI with a "Button Set"  I generally put a frame around the button set and then but a title in the frame like "Add" which then pertains to the entire set in the frame. Users seem to like that approach.

 

I was thinking this myself....  having 'Add' being within the 'Device' button logically separates it from the 'Scene' button. A frame around the button set with an 'Add' label would solve this - along with removing the 'Add' from within the Device button.

Link to comment

Hi Ron,

 

When you start typing, it should bring you a dropdown with scenes that match what you type. Or alternatively, if you the field is empty and you click on the arrow, you should see all your scenes in the dropdown.

 

If that's now what is happening, I'm starting to think that it may be a browser specific issue. What browser are you using?

 

Benoit

 

It does all of that... The only problem is not all my scenes are there...

I tries IE, Edge, Chrome, same with all  three, two different computers.

 

 

 

Ron

 

Link to comment

Junky,

 

I think we are alone... The thread has gone dead  

 

No, not alone! :-)

 

Hi guys.

Michel and Benoit, do you have any updated info for us concerning the status of the isy as a connected home device? It sure seems like it's taking a long time for Amazon to approve this. I realize that we just had a holiday though.

Any info would be appreciated.

Thanks

Junky

 

Hi Junky,

 

We had news today from the Amazon Echo Sr product manager. Amazon is about to have a new self-publishing method for connected home adapters and want us to go through it. That was supposed to be ready in late december, but the dev team is late. We are told to wait a "couple more days".

 

Benoit.

Link to comment

Benoit,

 

I have not yet looked at the Portal.  When I have a UI with a "Button Set"  I generally put a frame around the button set and then put a title in the frame like "Add" which then pertains to the entire set in the frame. Users seem to like that approach.

 

Hi Barry,

 

I'm currently using standard jQueryUI buttonset().

 

Would you have a screen-shot of what you suggest? That would help for inspiration to customize the buttonset.

 

You can PM me.

 

Thanks,

 

Benoit.

Link to comment

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...