This is the forum archive of Homey. For more information about Homey, visit the Official Homey website.

The Homey Community has been moved to https://community.athom.com.

This forum is now read-only for archive purposes.
GH

Connect Google home to homey?

edited April 2018 in Questions & Help
hi,

As we all know the homey speech recognition is bad. I had the chance to borrow a Google home for a few days and the speech recognition is really nice. But the support to connect to different hardware sucks. So it would be great to speech to the Google home device and send the certain commands to homey..

Would that be possible?

Better would of course that homey makes the speech recognition much better but I don't see this happening. There is no improvement at all since months. 
Tagged:
«134567

Comments

  • I saw something past week that Google Home has an open API, so maybe it is possible to build an app reading the API and parse the commands to usefull commands for Homey. Would be cool!
  • I would love this with either Google Home or Alexa (Dot). Not only the fact that we would have better speech recognition, but also speech in multiple rooms for cheap(er).

    I haven't looked into Home's API, but I think it'll be difficult with Alexa's API. Still; I would love it.

    Simply being able to send everything Google Home or Alexa hears to homey (perhaps even to a better logic variable) so it can be parsed would be awesome.
  • I can confirm that it is a lot(!) better with Alexa & IFTTT & Homey. I have to say "Alexa trigger [Command]" which is a bit more complicated but the recognition works like a charm. But you need the double effort to set it up because you need 1x flow and 1x IFTTT Applet for each command. I would appreciate if there would be a direct solution (without ifttt) like a Alexa Skill, or in this case a Google "Conversation Action" (Name of their "flow"/"skill") for Homey.  

    Google Home will have IFTTT aswell so you can connect it to Homey anyway.
  • Received Google Home yesterday, and the voice recognition is better than on the Google Pixel (Which also runs the new Google Assistant). I linked up Google Assistant to Homey with IFTTT, before I got the Home, to be able to use Assistant to trigger flows on Homey. So once I plugged the Home in, everything was already setup, and working like a charm.

    The two (Home and Homey) really compliment each other very well, since Home will probably never support the vast amount of hardware interfaces Homey has, as well as the easy Flow building interface, and Homey would probably never have the Voice recognition, search experience and massive amount of data google has (on us :P) It really made my Homey worth so much more.
  • @Weezul Thanks for sharing your experience. Do you need to set up each trigger word you want with IFTTT, or did you find a more elegant solution?
  • This sounds great! Is thus version of Google home already more languages supported?
  • WeezulWeezul Member
    edited December 2016
    @Weezul Thanks for sharing your experience. Do you need to set up each trigger word you want with IFTTT, or did you find a more elegant solution?
    No, unfortunately you have to set up each flow seperately. In IFTTT you have three possible sentences that will cause the Assistant to either make Homey say something, or trigger a flow. There are options for variables, which I assume can be passed on to Homey, which you can maybe use there. I am not sure if that is possible.

    I have a few scenes ('cooking', 'eating', 'watching tv', 'going to bed', 'normal lighting') which for each I made an IFTTT trigger, running on a few sentences. It would have been better if it was possible to just do trigger words, instead of sentences that have to match exactly.

    That said.... The convenience of saying "Hey Google" followed by either: "We are off to bed / I am off to bed / We will go to bed now", and it having the TV, lights and heating turned off plus bathroom light with the bedroom nightstand on is for me already quite satisfying :smiley: 

    Another thing I would have like to have, is more control over Home... like sending it commands from Homey for instance, since it does not have any proper presence system, and does not do things without being manually triggered. Unlike Homey that can tell my girlfriend when I left work (Using Tasker).

    EDIT: I briefly looked at Google Actions yesterday, it looks promising, but I would have to spend more time on it to see if we can use it do integrate Home and Homey better
  • @Weezul yep, I did exactly the same. Linked my Pixel to IFTTT / homey. Takes a bit to set up, but once that is done it works like a charm. What's this google actions you're talking about? 
  • PeaoPeao Member
    edited December 2016
    Is there a cheap (<180€) way to get Google Home in EU, yet? I can only find 250€+++ auctions. I am pretty happy with Echo and Homey but I think Home will fill this spot better than Echo, since I've got 8 Chromecast (Audio and Regular) in my setup.
  • WeezulWeezul Member
    edited December 2016
    @EvertorN, you can take a look at it here: https://developers.google.com/actions/

    It is meant more for mobile/web-app developers, so not sure if we can use it directly like that, or if we would have to make a dummy app to run on the phone, which we would then have to tell Google Home "Hey google, let me talk to Dummy App"... which would be a pity. 

    @Peao: 8 Chromecasts!!! Get Home and you would have 9 :)
    I got mine via a friend that was in the US, at the moment I think that is the only reasonably priced way :(
  • Something I found out because I was interested (and also posted on another thread in this forum, but just for people who are interested):

    If some developer has access to a dot/echo, there should be a way to get Alexa set up in a way so that the following works: "Alexa, tell Homey to turn on the lights", where "turn on the lights" is send as a string to homey via HTTP commands (i.e. via betterLogic).

    To help a developer a bit on the way, this is what I found till now:
    https://github.com/ekt1701/Alexa-Wolfram-Alpha --> An Alexa-Wolfram app which works similarly, in that you say "Alexa, ask Wolfram Alpha what the value of pi is", where "what the value of pi is" is send to wolfram-alpha, and the response is returned.

    Appearantly, it comes down to the following code:

    query = intent['slots']['response'].get('value') 
    query = query.replace(' ', '%20') 
     url = "http://api.wolframalpha.com/v2/query?podindex=2&format=plaintext&appid=" + app_id +"&input="+query


    However, a big downside is that this works via the LITERAL slot type reference. If I understand Alexa's documentation correctly (https://developer.amazon.com/public/solutions/alexa/alexa-skills-kit/docs/alexa-skills-kit-interaction-model-reference#literal-slot-type-reference), the LITERAL slot type reference is planned to stop working in roughly a month's time. I do not know if this will actually make all the LITERAL slot types stop working, or that the skill using the LITERAL slot type will not get accepted by Amazon.
    If it's the latter, a Homey app can technically still be made, but just for personal use.
  • This is awesome news. Athom will ofcourse never be able to match Google in terms of the work they can put into speech recognition. If the Google assistant can be integrated well with Homey it will be fantastic! Right now im excited about the newly announced Nvidia Spot as I already have 2 Nvidia Shields. If you could communicate with Homey by talking to the Nvidia spot and Google assistant it would open up a whole new set of possibilities

    /Jacob 
  • Weezul said:
    @Weezul Thanks for sharing your experience. Do you need to set up each trigger word you want with IFTTT, or did you find a more elegant solution?
    No, unfortunately you have to set up each flow seperately. In IFTTT you have three possible sentences that will cause the Assistant to either make Homey say something, or trigger a flow. There are options for variables, which I assume can be passed on to Homey, which you can maybe use there. I am not sure if that is possible.

    I have a few scenes ('cooking', 'eating', 'watching tv', 'going to bed', 'normal lighting') which for each I made an IFTTT trigger, running on a few sentences. It would have been better if it was possible to just do trigger words, instead of sentences that have to match exactly.

    That said.... The convenience of saying "Hey Google" followed by either: "We are off to bed / I am off to bed / We will go to bed now", and it having the TV, lights and heating turned off plus bathroom light with the bedroom nightstand on is for me already quite satisfying :smiley: 

    Another thing I would have like to have, is more control over Home... like sending it commands from Homey for instance, since it does not have any proper presence system, and does not do things without being manually triggered. Unlike Homey that can tell my girlfriend when I left work (Using Tasker).

    EDIT: I briefly looked at Google Actions yesterday, it looks promising, but I would have to spend more time on it to see if we can use it do integrate Home and Homey better
    I realy don't see the issue here: "That said.... The convenience of saying "Hey Google" followed by either: "We are off to bed / I am off to bed / We will go to bed now", and it having the TV, lights and heating turned off plus bathroom light with the bedroom nightstand on is for me already quite satisfying smiley "

    I use this already for a long time now: (Okay Homey .. i ga slapen) and preforms very well (very rarely I have to say it twice, (and mostly at times the screensaver has been switched off (every night at 22?:00) (switched on at 07:00) ...
    Did you try Dutch ?
  • Hi there - interesting topic to follow so I'm dropping in if you don't mind :)

    I'm a new user (really new, new - using Homey since yesterday). I definitely love the ease-of-use and polished interace. The product look promising but I'm really disappointed about voice accuracy. I've mostly fitted my house with zwave components and Homey's voice was one of the reasons to buy Homey versus a combination between e.g. Alexa & Smartthings.

    Regarding the below par voice experience - I already created a support ticket and they actually made me aware of this forums so I've just been browsing around to learn more. One of the scenarios I'm thinking of is using Alexa / Google Home through IFTTT to combine best of both worlds like other users in this topic already have been doing.

    However - I'm not sure if I'll get to a great product experience because I'm kinda scared for an extensive delay between IFTTT&Alexa/GHome and Homey triggering a flow. Could some folks in this topic with either a Google Home/Alexa share their experiences with IFTTT in combination with Homey? I'm especially interested about the responsiveness.

    Thank you very much!
  • bvdbosbvdbos Member
    edited January 2017
    @lenonline There's a lot of postings regarding voice on Homey. One of the most convenient things is
    http://[homey-ip]/manager/speech-input
    Also, with the better voice app you can improve Homey understanding you.

    I don't use IFTTT but I understand it's really slow...
  • lenonlinelenonline Member
    edited January 2017
    @BasVanDenBosch: thanks man! Annemarie indeed forwarded me to the same place. I've been using it to 'troubleshoot' the speech issue. It also got me thinking and evaluating other options; simply because Homey doesn't correctly pick up any command other than 'whats the time' and 'what's the date'. I've already spend a few hours tailoring the speech commands without much luck. Distance between Homey and me is about 2 metres.

    Then again, the mic in my smartphone is (almost) failproof which is absolutely fabulous. Using your phone, you really experience the power of Homey's concept. However, I didn't buy a Homey to then take out my phone every time when I have a request, haha.

    Now I wish that I would have discovered the forum before buying; maybe I'm the one to blame for making the wrong assumption that Homey is production ready because it's available at MediaMarkt (and other online retailers).. I don't know :(. Of course, I can return Homey back but I really do like the quality of (most of the) engineering plus the concept/vision including the idea of supporting a Dutch startup. However, what I really don't like, is the lack of a outlook on how Athom is going (to try) to solve these issues; it seems they are focused on other issues on the roadmap and that really makes me doubt if I'm in the right ecosystem.. #hencegooglehome #oramazonecho ;)
  • I too would be interested in using something like the Nvidia Spot (running Google Assistant on my Nvidia Shield, if possible with a link to Homey), but then, I do feel it's kind of sad that we have to resort to this kind of options instead of receiving a clear reply from Athom in regard to what we can expect in terms of voice recognition of Homey. I really wish @Emile would be clear if we can still expect any improvements, and if so, how much. It's been ages since we've received the last official comment related to the quality of voice recognition.
  • It will never be as good as Google home. I have no faith in this anymore. And yes communication of athom is one of my points to say this. 
  • BumblezBumblez Member
    edited January 2017
    I don't think the (hardware) quality of the microphones is the major issue here... I believe it's all in the software (translating the sound sample from the microphone into a text). So why don't they just take a license of Google's software algorithm for that? That should make Homey's voice recognition on par with that of Google. I'm sure that's what Nvidia does too, with their Spot...
    On one hand, yes, it will be a significant investment, but on the other hand, we're talking about one of the major features of Homey here... if this doesn't work well (especially for the retail customer, at which stage Homey is now), it can well be Athom's downfall. And even for the people who already have a Homey... we already see a lot of people adding (or considering) alternatives for voice recognition... those setups can easily turn into situations where Homey is completely obsolete.
    In short: I believe that the investment to implement Google's voice recognition software in Homey, would probably solve the problem and thus earn itself back over time.

    [Edit] on top of that, apparently Google's speech recognition already supports 80 languages, meaning it could give Athom's language support a huge boost, and save them significant money (and time) in the multi-language support field.
  • Bumblez said:
    So why don't they just take a license of Google's software algorithm for that? That should make Homey's voice recognition on par with that of Google. I'm sure that's what Nvidia does too, with their Spot...
    I think this is not the way to go, simply because Google will charge a monthly payment instead of a one-time payment. This would mean that in the end, either end-users have to start paying monthly fees, or Athom needs to get monthly money from us (the users) in another way, for example payment for apps for which Athom gets a cut.

    I severely dislike both options.
  • That depends on how much it would cost... apparently they have special prices/conditions for "embedded devices (e.g., cars, TVs, appliances, or speakers)", as they ask developers of those to contact them directly. So if for Homey, for example, the cost would come down to say 1000.- overall per year, then Athom could decide to just accept these costs and consider this "included in the price of Homey"... no need to turn this into a monthly charge for the users (you aren't charged a monthly fee for Nvidia Spot's voice recognition, either).

    Now they already did a costly speech implementation for Dutch/English languages, that still isn't working very well. Fixing that will also cost a significant amount of time and money, and adding other languages will cost even more.
    So while using Google would add some costs, it would reduce some others...
  • I've to agree with @Bumblez. And on top of that, the lack of transparancy and communication from Athom on this worries me most. Earlier I read a nicely written summary listing the challenges by an Athom developer. He mentioned the differences between Amazons/Google's and Athom's development teams; especially the difference in size. That's okay - however - if your vision is 'voiced controlled homes' than I don't buy into this argument.

    If you can't get it to work and you lack the resources, get your act together and change your strategy. You will probably piss off people but I'm convinced that this type of honesty will get you more customers.. How can you expect me to recommend Homey to less tech-savy friends and colleagues? I would love to though!
  • @lenonline There's a lot of postings regarding voice on Homey. One of the most convenient things is
    http://[homey-ip]/manager/speech-input
    Also, with the better voice app you can improve Homey understanding you.

    I don't use IFTTT but I understand it's really slow...
    I use alexa / IFTTT/Homey  and it is twice as fast between talking to alexa and Homey doing the action, than using homey to recognise the voice command (even if it does it first time, which it raely does)
    Homey's voice recognition is not only very bad in english, its very very slow as to be unusable.
  • Jeroenski74Jeroenski74 Member
    edited February 2017
    Peao said:
    I can confirm that it is a lot(!) better with Alexa & IFTTT & Homey. I have to say "Alexa trigger [Command]" which is a bit more complicated but the recognition works like a charm. But you need the double effort to set it up because you need 1x flow and 1x IFTTT Applet for each command. I would appreciate if there would be a direct solution (without ifttt) like a Alexa Skill, or in this case a Google "Conversation Action" (Name of their "flow"/"skill") for Homey.  

    Google Home will have IFTTT aswell so you can connect it to Homey anyway.
    You own quote ;-)
  • Jeroenski74Jeroenski74 Member
    edited February 2017
    I was able to test the interface Google Home -> IFTTT -> Homey without a real Google Home device:

    - Create new agent on api.ai (following the sample of google) called Homey with a default intent:


    - Made one entity (Command) don't know if this is really needed, probably not but it is working and you can enter as much command as you want to be understood by Google Home (This way you get immediately feedback of Google Home if something is said that shouldn't be understood, meaning that it has no defined flow on Homey)


    - Use the Google Assistant integration:


    - Made a IFTTT applet responding on Google assistant triggered by the word Homey followed by a command



    - IFTTT triggers the flow GoogleSays
    - Flow GoogleSays will execute emulate speech as if you are talking to Homey:



    And it works perfectly. Now I have to find me a Google Home for a nice price!

    Happy connecting Google Home and Homey :smile:
  • JesperJesper Member
    edited February 2017
    @Jeroenski74; How fast does Homey react to the input via Google Assistant (Google Home)? Maybe you can post a video of it on this forum  :). I'm asking this because a lot of people are worrying about the delay because you have to use IFTTT.
  • Here a movie which shows adding a new command and the delay via IFTTT (I'm not a vlogger actually this is my first upload ever, so don't judge it to hard  ;) ) :



  • New implementation of flow with http cards to connect Google Home directly to Homey. See my previous post using ifttt to setup Google Home (simulation)

    Add fulfillment:


    Speech will contain any command you want to give (only english is understood well by Google Home currently, but you can simulate with Dutch text as if it is correctly understood)

    In homey you need to install http and the following two flows, first one to extract the command:



    The JSONPath looked for is $.result.fulfillment.messages[0].speech which will be sent to the trigger GoogleSays (HTTP trigger)

    The trigger will start the next flow:



    which in turn will emulate the command that comes with the trigger. By emulating the command it is as if Homey is hearing the command itself and will start ALL your other already existing flows triggered by Ok Homey.

    This works much faster (almost instantaneous) than ifttt. Skipping ifttt will also make many people happy (I hope).

    Good luck, if you need extra help just send me an pm.
  • Werkt overigens ook met een Nederlands versie: 'zeg tegen Jarvis' vertel eens een grap
  • @Jeroenski74 To be clear: Can this be set up in a way that everything you say to Google Home is send directly to Homey via just HTTP? That would be amazing.

    I mean this would work as "Google, turn on the lights", rather than "Google, ask Homey to turn on the lights". 
    Can you confirm/deny this?
Sign In or Register to comment.