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.
Closed Solved

Homey understood me, but "she is not sure what she would like me to do"

MarkVMarkV Member
edited September 2017 in Archive
So speech recognition is not prefect in Homey. Since you can figure out what she hears, it's doable to set up a set of phrases which she will see as the right phrase. (drag and drop multiple green text labels get's slower for every new label which is another bug). 

But if you do get it set up, see will both understand and not understand you at the same time.

So if you set up Homey as attached:

She will both say "see you later" and " I'm sorry, but I am not sure what you would like me to do". 




Modbreak:
Removed the bug sentence from the topic title because it's not a bug but a change from March 2017 in the Homey core

Tagged:

Comments

  • lubbertkramerlubbertkramer Member
    edited September 2017
    It's not a bug you are using the cards wrong, since 1.2.2 this has changed and the way you use it is deprecated . You need to use the microphone card in the IF and insert there what has been said. There are multiple topics about to find on the forum for example here -> https://forum.athom.com/discussion/3089/homey-im-sorry-but-im-not-sure-what-you-would-like-me-to-do

    Or see the update from Athom -> https://forum.athom.com/discussion/comment/49280/#Comment_49280 It has also been showed in the "whats new" screens you get after updating

    Note: as of Homey v1.2.0, a much requested feature was feedback when Homey did not understand what you said. That now works, with the small drawback that it requires everyone to migrate their 'When Homey said something And the text contains' cards to the 'When something specific was said' card, if that hasn't happen yet.

  • It's not a bug you are using the cards wrong, since 1.2.2 this has changed and the way you use it is deprecated . You need to use the microphone card in the IF and insert there what has been said. There are multiple topics about to find on the forum for example here -> https://forum.athom.com/discussion/3089/homey-im-sorry-but-im-not-sure-what-you-would-like-me-to-do

    Or see the update from Athom -> https://forum.athom.com/discussion/comment/49280/#Comment_49280 It has also been showed in the "whats new" screens you get after updating

    Note: as of Homey v1.2.0, a much requested feature was feedback when Homey did not understand what you said. That now works, with the small drawback that it requires everyone to migrate their 'When Homey said something And the text contains' cards to the 'When something specific was said' card, if that hasn't happen yet.



    Something like this?
    flow.png 756.6K
  • lubbertkramerlubbertkramer Member
    edited September 2017
    MarkV said:
    It's not a bug you are using the cards wrong, since 1.2.2 this has changed and the way you use it is deprecated . You need to use the microphone card in the IF and insert there what has been said. There are multiple topics about to find on the forum for example here -> https://forum.athom.com/discussion/3089/homey-im-sorry-but-im-not-sure-what-you-would-like-me-to-do

    Or see the update from Athom -> https://forum.athom.com/discussion/comment/49280/#Comment_49280 It has also been showed in the "whats new" screens you get after updating

    Note: as of Homey v1.2.0, a much requested feature was feedback when Homey did not understand what you said. That now works, with the small drawback that it requires everyone to migrate their 'When Homey said something And the text contains' cards to the 'When something specific was said' card, if that hasn't happen yet.



    Something like this?
    No because you still use the contains card in the AND part, the microphone card has a card "where user said", the contains card may not be used with speech and will result in the "sorry i didn't understand"" but execute the flow
  • Thanks for you help. Still no sugar.

    I would like homey to respond to: 
    go to sleep
    go to the
    go to see
    goto sleep
    goto the
    goto see

    Because that what she hears.

    The test flow below did not work:



    Even if the output exactly matches "go to sleep', she won't understand what I want her to do. The word 'Sleep' alone works but only if the hears that exact work right. 

    How do I get her to listen to:
    go to sleep
    go to the
    go to see
    goto sleep
    goto the
    goto see

    Like she did in the opening post flow?
  • lubbertkramerlubbertkramer Member
    edited October 2017
    MarkV said:
    Thanks for you help. Still no sugar.

    I would like homey to respond to: 
    go to sleep
    go to the
    go to see
    goto sleep
    goto the
    goto see

    Because that what she hears.

    The test flow below did not work:



    Even if the output exactly matches "go to sleep', she won't understand what I want her to do. The word 'Sleep' alone works but only if the hears that exact work right. 

    How do I get her to listen to:
    go to sleep
    go to the
    go to see
    goto sleep
    goto the
    goto see

    Like she did in the opening post flow?
    You can only use one microphone card in the IF 
  • Ok, I  tried doing it like this https://forum.athom.com/discussion/comment/51963/#Comment_51963 (which is how it now should work) and Homey fails miserably all the time. I even had native English speaker in the house an no sugar.

    The original flow from 1.2 https://forum.athom.com/uploads/editor/lr/637f1mhqufd5.png did work because it compensates for homey being a bad listener, so I went back to that and now enjoy the extra "not sure what she would like me to do" messages. 
  • @MarkV
    install better voice because don't know if this still works with latest firmware but don't get your hopes up that  athom will improve speech recognition it will only get worse due to the forbidden words lists they are expanding without giving the user an opportunity to disable it. Like everything off  
  • MarkVMarkV Member
    edited October 2017
    This result in a "i am not sure what you would like me to do" and no action. (where the /manager/speech-input/ does detect the right output)
This discussion has been closed.