Jump to content

Interface with Amazon Echo?


awzulich

Recommended Posts

Benoit,

 

Yes you are correct a scene does not have a state, only the devices therein have states.  IIRC that is why a scene can't be dimmed.

 

My statement about comma and period in text representing speech is just a comment.  The same is true when reading text aloud to a group. 

 

Barry

 

P.S. your presence on the forums is greatly appreciated, as is Michel's. too many forums, other than UDI, do not have a visible presence from their developers / decision makers. 

Link to comment

My statement about comma and period in text representing speech is just a comment.  The same is true when reading text aloud to a group. 

 

I wonder if this could be used with the skill in the spoken phrase section of notes?  I have more than one device in almost every room and so to address them I'm unable to use e.g., family room light / family room lamp.  If adding a comma indicated more were to follow and period the end then this might allow an extended phrase with the same heading.

 

"or just quotes"  :)

 

 

Jon...

Edited by jerlands
Link to comment

I wonder if this could be used with the skill in the spoken phrase section of notes?  I have more than one device in almost every room and so to address them I'm unable to use e.g., family room light / family room lamp.  If adding a comma indicated more were to follow and period the end then this might allow an extended phrase with the same heading.

 

"or just quotes"  :)

 

 

Jon...

 

Unfortunately that would not work. Alexa will never come up with a comma or a dot in it's speech recognition.

Link to comment

Benoit,

 

Yes you are correct a scene does not have a state, only the devices therein have states.  IIRC that is why a scene can't be dimmed.

 

My statement about comma and period in text representing speech is just a comment.  The same is true when reading text aloud to a group. 

 

Barry

 

P.S. your presence on the forums is greatly appreciated, as is Michel's. too many forums, other than UDI, do not have a visible presence from their developers / decision makers. 

 

Actually, a scene can be dimmed. /rest/nodes/<device>/cmd/BRT and DIM will work on a scene. However, we can't set a scene to an absolute level. /rest/nodes/<device>/cmd/DON/<level> will turn on devices in the scene to their on level, not the level specified in the url.

 

Interestingly, a zwave device in a scene will respond correctly to an absolute level!

 

Thanks for your kind comments,

 

Benoit.

Link to comment

Unfortunately that would not work. Alexa will never come up with a comma or a dot in it's speech recognition.

 

Well how 'bout quotes then :)  Anyway.. when the Connected Home certification (or whatever it is) passes hopefully then we'll have longer phrases and better understanding.

 

 

Jon...

Link to comment

Does anybody have two word names working?  For me, I need to fill in the spoken word in the notes screen but only single words work.  Kitchen works, but Kitchen Lights or Kitchen Pendants does not.... nothing with two words.  Does anyone have programs working?  If so, what are they called?  I get a variety of different issues depending on the program name.

Link to comment

In addition to Michel's suggestion, if you have not changed anything on your ISY, you may want to try a reboot.

 

Alternatively, you could try the following.

Telnet to your ISY

Command SPU will allow you to change your portal URL. Change it to something else, then bring it back to https://my.isy.io/dispatcher

 

Benoit.

 

Tried an ISY reboot. Opened a Telnet session, but "SPU" doesn't accomplish anything.

Link to comment

Does anybody have two word names working?  For me, I need to fill in the spoken word in the notes screen but only single words work.  Kitchen works, but Kitchen Lights or Kitchen Pendants does not.... nothing with two words.  Does anyone have programs working?  If so, what are they called?  I get a variety of different issues depending on the program name.

 

I've experimented with IFTTT and am able to say "Alexa, trigger family room light on" / "Alexa, trigger family room light off" / "Alexa, trigger family room lamp on" etc., I don't know the limit on the length of the phrase but for now four words e.g., "family room light off" has been enough.

 

 

Jon...

Link to comment

Hi stusviews,

 

Please go to Admin Console | Configuration | Portal and make sure it's still registered.

 

Once that's done, please check your error log for any connectivity issues. If you wish, you can send your error log to support@universal-devices.com.

 

With kind regards,

Michel

 

Oops, wrong log sent. Correct log is on the way.

Link to comment

I have also spent quite a bit of time over the last two days playing with both the UDI skill and other Alexa integrations(skills) mainly the IFTTT. Obviously Alexa can handle more complex phrases as it does a very good job in IFTTT. I have several recipes that include 4 or 5 word phrases and elexa has about a 99% accuracy with understanding what I am asking for. The other thing is with IFTTT you can type in any phrase when you are creating a recipe and it is able to understand it.

 

I fully appreciate the effort that UDI is putting into these integrations but if it's release was a bit premature (because the second certification will make it better) maybe it's launch should be reconsidered? Again, don't get me wrong I think we all appreciate the effort but I personally would rather see the engineers and developers spending time making the skill rock solid vs. trying to get this version sort of working!

 

I can't say thank you enough UDI, this stuff really is cutting edge! Keep up the great work with whatever direction you decide!

 

Thanks

Jim

Link to comment

I think the reason IFTTT matches well is that it's only trying to match against full phrases that you have defined, rather than breaking your speech into separate words and trying to determine what phrase was spoken from separate words. IFTTT with its whole phrase machine will always be more accurate IMO. However, you have to define each specific full phrase. So 'trigger turn on family light' is very different to 'trigger family light on'.

 

They are different solutions with different benefits and limitations.

Link to comment

I just played with it for about an hour.  It recognizes what I am saying just fine according to the app.  I still have to say Alexa open Izzy and then say Alexa tell Izzy to switch "X" on or off etc.  Then I have to start all over and open Izzy or it wont work.

 

 I will ask some friends at work that are using Alexa with Homeseer how its working for them tomorrow.  I am guessing that Alexa is still being fine tuned. 

 

Hopefully the next version will be better.

Link to comment

Hello to Michel and all,

 

I got my Echo to work with my ISY today.

 

FYI: my testing so far.

 

One of the things that I discovered is that setting a temperature set point is in error. If I want my Heating Set point to be 70 degrees, I have to say to set the temperature set point to 140 degrees, exactly double. I am using  Venstar Thermostats one with the add-on insteon module and the other with the built-in module. The one with Built-in module does not respond and the other one does. It definitely reads the temperature back accurately (not double). 

I find that Alexa responds to single word names and not all of them. for instance it doesn't understand the word closet. I also noted that even though I labeled my Bathroom switches Bathroom 1 and Bathroom 2, Alexa dose understand the word Bathroom and not the real names. I have NOT had success with multi-word names.

 

Question:

is there anything that we supporters can do to help enhance the ISY to Echo experience? How can we support UDI in this effort?

 

And finally:

Thanks UDI and Michele for all the work that you have done to get us this far.

Edited by C Martin
Link to comment

I think the reason IFTTT matches well is that it's only trying to match against full phrases that you have defined, rather than breaking your speech into separate words and trying to determine what phrase was spoken from separate words. IFTTT with its whole phrase machine will always be more accurate IMO. However, you have to define each specific full phrase. So 'trigger turn on family light' is very different to 'trigger family light on'.

 

They are different solutions with different benefits and limitations.

 

Yes, IFTTT creates a phrase and I'd like like to see the 'spoken field' (device>notes>spoken) also become a phrase rather than parsed words.  That would allow me more than two words to describe the device and I'd be able to use the same heading more than just once only with a different ending..  problem is.. what did I name that device :)

 

 

Jon...

Link to comment

Hi,

 

I use and enjoy my ISY every day.  Just installed ISY Portal and linked to Amazon Echo Alexa skill.

 

Some observations and questions:

 

1) Same problem mentioned above: Single word vs Multi (space) words  - Izzy turn on Living Room Light fails, "what device"- Rename Living Room to Main and Izzy turn on Main Light works = also renaming to LR, Izzy turn on "el" "ar" light works.  Echo sometimes stumbles on Downstairs - Echo says "I can't find device "stairs""

 

2) Does Echo/Izzy understand/use SCENES?  I have scenes with names like Bedroom that turn on a couple of insteon devices - Izzy does not seem to ever respond to Bedroom, only individual devices in the Bedroom scene.  It would be great to have IZZY work with scenes

    If scenes aren't supported - please give some generic examples of  "workaround" programs that IZZY can use to turn on/off multiple devices and not require total re-write of existing programs.

 

3) Thanks for creating this sub forum just for IZZY - how about some stickies at the top with IZZY directions/setup/syntax for reference/noobies

 

 

I have confidence this will be improved.

Link to comment

Hi Jim,

I fully appreciate the effort that UDI is putting into these integrations but if it's release was a bit premature (because the second certification will make it better) maybe it's launch should be reconsidered? Again, don't get me wrong I think we all appreciate the effort but I personally would rather see the engineers and developers spending time making the skill rock solid vs. trying to get this version sort of working!

I am not sure how you got the impression that this is a production ready solution. It's not and we never claimed it to be (otherwise it would have been advertised all over the place). Furthermore - and explicitly - I stated that there are some issues and that you are welcome to "test":

http://forum.universal-devices.com/topic/17082-interface-with-amazon-echo/?p=157836

 

 

Hi whywork,

 

Yes, scenes should work. The issue is with the words used such as bedroom and light.

 

We are definitely working on improvements. IFTTT is not using a skill otherwise you would have to have: "Alexa, tell IFTTTT to trigger xxx". The issue is with Skills and slots that may have other meanings to Alexa. Personally, I think the concept of skills and predefined slots is not going to work for automation because there's really no way to have all different permutations and combinations of locations predefined. We are waiting to hear back from Amazon and hopefully we can figure out a way to make this a little less stringent. In the meantime, Benoit has been playing with utterances and interpretations by Alexa but with very little improvements: Alexa needs a list of words/phrases from which it can infer what was said based on some weights for each.

 

With kind regards,

Michel

Link to comment

Maybe the Connected Home is the better direction for automation.

If Amazon added a few verbs in addition to turn on and off if would likely do 99% of what is commonly used.

In practice  Alexa tell Izzy to turn on..   is a little cumbersome vs Alexa turn on...

 

Having the ISY portal available to those that want to create a skill is tremendous.

 

(Now Alexa just needs to respond to "Clap On"  "Clap Off".  :-P )

Link to comment

The portal still indicates that the ISY has been offline since Sunday, December 13th, 2:11:41PM. I tried deleting and reinstalling the ISY from the portal. That didn't help at all. Do I give up on being able to use Echo? Is there a refund procedure so I can begin again? Is there no solution?

Link to comment

I do not believe the portal is necessary if you want to create your own skill. The Amazon cloud will send the slot information that it has decoded from the speech it was sent by the Echo to an endpoint you define. You can host the skill's endpoint on your LAN as long as you have a certificate and can handle SSL (I use sTunnel on an RPi to handle the SSL issues and the certificate).  Your endpoint can then deal with the ISY using the ISY restful interface.  The Skill endpoint could even subscribe to ISY events and know at all times what the status of every device in the ISY configuration is.

 

I have a basic skill named "Sarah" running on a LAN local endpoint and was going to start developing the necessary utterances and endpoints, but based on the difficulty the UDI team seems to be having with all the links in the process chain I am going to stay with my Connected Home and the HA Bridge (HUE emulator) for now.

 

One thing I believe is that Alexa gets "Smarter" as she hears the same phrases over and over.  For example the Echo rarely misses "Kitchen Lights", "Master Bedroom Lights" and other devices which I command often. Things I command less e.g. "Guest bedroom Lights" have a lower recognition rate until I take the time to speak them multiple times. I am not sure how this happens but I believe it is so although I have run no definitive test.  The Echo seems to have no issue with multiple words as the friendly name e.g. "Office Fan Medium" in the connected home scenario. Also there seems to be reasonable flexibility in phrasing the command (turn on, turn off, shut, shut off  . . .  with or without a "the" before the device name. One of my biggest problems is I forget the names of the 90+ devices I have setup in the HA Bridge. By convention I have lights always as "<room name> Lights" not "<room name> Light". Lamps are always referred to in the singular e.g "Barrys Lamp" or "Guest Bedroom Lamp". I use the suffix "Area" as in Kitchen Area" to designate a scene that handles multiple devices. Works well for me with my aging memory.

 

Benoit has already indicated that I should expect the same level of performance when the UDI ISY Connected Home is certified and released.

 

Barry

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...