Announcement

Collapse
No announcement yet.

Kinect Plug-in - Feature Requests

Collapse
This is a sticky topic.
X
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    #16
    How about linking a Kinect with a Speaker client? Im going to use Kinects in place of the Mics for a Speaker client, so directed output would be cool.

    Comment


      #17
      Originally posted by surovich View Post
      How about linking a Kinect with a Speaker client? Im going to use Kinects in place of the Mics for a Speaker client, so directed output would be cool.
      What do you mean exactly?
      The speaker client doesn't have any external API, so it's hard to link anything to it.

      Comment


        #18
        Kinect instance name

        Hi Spud,

        A couple of suggestions:

        Set the name of the Kinect that received the command on a global variable, so that I can use the same command in different rooms to enable the action in that room only (i.e. Play music will turn on the music in the room I am in, rather than having to say Play music in living room)

        Able to support multiple Kinect languages at once. I have set up some phrases and by default the Kinect language is English. My wife (native English speaker) consistently gets 98% recognition while I (native French speaker) get around 65 - 70%. When I set the Kinect language to French I then get 98% and she gets around 45%. The funny thing is that my phrases are still in English...Kinect just recognizes my French accent better I thought of trying with a French Canadian language to see if Kinect will understand both of us reliably. I know I can reduce the confidence threshold but then I get a lot of false positive.

        Thanks,
        Eric

        Comment


          #19
          Originally posted by Nan0Tr3k View Post
          Set the name of the Kinect that received the command on a global variable, so that I can use the same command in different rooms to enable the action in that room only (i.e. Play music will turn on the music in the room I am in, rather than having to say Play music in living room)
          I assume you have one kinect in each room so, you use one instance of the plugin for each room. As each instance is independent and has its own trigger category ("Kinect Trigger - instance_name"), you have to create a different trigger for each room. So you can already know which Kinect has received the command if you create one "Play music" event per room.

          Originally posted by Nan0Tr3k View Post
          Able to support multiple Kinect languages at once. I have set up some phrases and by default the Kinect language is English. My wife (native English speaker) consistently gets 98% recognition while I (native French speaker) get around 65 - 70%. When I set the Kinect language to French I then get 98% and she gets around 45%. The funny thing is that my phrases are still in English...Kinect just recognizes my French accent better I thought of trying with a French Canadian language to see if Kinect will understand both of us reliably. I know I can reduce the confidence threshold but then I get a lot of false positive.
          I know what you mean, I'm a native French speaker too, and I always have a hard time when I try to test with the English language.
          I don't know if it is possible to load 2 different recognizer at the same time. Even if it's possible it would be difficult to deal with, as both recognizers could recognize the same command or different ones at the same time. Moreover this is not a feature that a lot of people would use, so I can't promise you anything.

          Comment


            #20
            Thanks for your reply. I was trying to avoid duplicating events, but that will work.

            Don't worry about the multi language. Do you know how far apart each Kinects need to be to avoid interference? I assume that this depends on the layout of the rooms, the position of the Kinects and reverberation.

            Comment


              #21
              Will there be or is there support for the new kinect v2?

              Comment


                #22
                yes I plan to add support for kinect v2
                but no ETA yet

                Comment


                  #23
                  Would it be possible to get person identification? It would be pretty big to have HS be able to know the identity of and count the occupants in the house.
                  Originally posted by rprade
                  There is no rhyme or reason to the anarchy a defective Z-Wave device can cause

                  Comment


                    #24
                    I have come across a little issue with the wife and the kinect VR. She really loves the setup but she forgets the commands that make things work LOL. I was thinking would it be possible to have a action that would have the plug in send a list of the commands available in the grammar for the plug in and have HS speak it ie. The available commands are <list from grammar> it would greatly increase the WAF

                    I have been thinking about just putting the commands in a text file and having HS read it but if there was something coded in the plugin it would be easier to keep updated because it could pull from the grammar database and always be up to date with the commands available


                    Cheers Ken
                    HSPRO,HS3PRO, BLsecurity, , MCSxap, HSTouch Android, UltraGCIR, CM15a, USBUIRT, BLUSBUIRT, WIFIRGB, BLAB8SS, BLcontrol, BLGData, BLLAN,BLOccupied, BLRadar, BLVolume, iTunesDAAP, UltraGCIR3, Airplayspeak, BLalarm, BLbackup, BLLED, BLrandom, BLReminders, BLRF, BL Speech, Hyperion, IFTTT, KINECT, XBMC, MCSprinkers PRO, PHLocation, ULtrapioneer, Ultralog, ultraweatherbug, Z,troller, GC-100, GC WIFICC, GC-WIFI SERIAL, Nitrogen logic depth camera controllers,

                    Comment


                      #25
                      Originally posted by S-F View Post
                      Would it be possible to get person identification? It would be pretty big to have HS be able to know the identity of and count the occupants in the house.
                      @ Spud,

                      Any thoughts on this?
                      Originally posted by rprade
                      There is no rhyme or reason to the anarchy a defective Z-Wave device can cause

                      Comment


                        #26
                        AEC and Suppression?

                        Any thoughts on implementing Automatic Echo Cancellation/Suppression into the plugin? Would be neat to have the plugin instance running on your HTPC for instance, so it can filter out any audio from whatever you are watching to have far more reliable VR.

                        Comment


                          #27
                          Originally posted by S-F View Post
                          @ Spud,

                          Any thoughts on this?
                          yeah person identification is something I want to add, but not sure how to implement it. My first idea is to identify someone using length of the detected body parts: arms, legs, upper body etc..., but I'm not sure how relialble it would be.

                          Another more complicated way is to use face tracking, but this would only work if you face the kinect sensor.

                          Comment


                            #28
                            Originally posted by nikku View Post
                            Any thoughts on implementing Automatic Echo Cancellation/Suppression into the plugin? Would be neat to have the plugin instance running on your HTPC for instance, so it can filter out any audio from whatever you are watching to have far more reliable VR.
                            yes this will be in the next version of the plugin

                            Comment


                              #29
                              Originally posted by spud View Post
                              yeah person identification is something I want to add, but not sure how to implement it. My first idea is to identify someone using length of the detected body parts: arms, legs, upper body etc..., but I'm not sure how relialble it would be.

                              Another more complicated way is to use face tracking, but this would only work if you face the kinect sensor.

                              How does the XBOX do it? Personally I have no experience with the XBOX at all. AT ALL. But I've heard that it can identify people. A friend told me that he played a game at a mutual friend's house with a Kinect once. Then several months later he went to this person's house while he was playing a game. The System recognized him and asked if he wanted to join the game.
                              Originally posted by rprade
                              There is no rhyme or reason to the anarchy a defective Z-Wave device can cause

                              Comment


                                #30
                                version 3.0.0.16

                                in version 3.0.0.16 available here, I have added echo cancellation

                                To enable it, go to the config page, in the Voice Recognition tab tick the "Echo Cancellation" checkbox then restart the plugin.

                                This will filter out all the audio from your default speakers (the audio output which is set as default in Windows)

                                To check that it works correctly set the log level to trace, then start a movie on the same computer that is running the plugin. If you don't have the "Echo Cancellation" ticked you should see in the logs that the plugin tries to understand the movie dialogues as commands, if you have it ticked there should be almost no log lines written as all the audio from the movie is filtered out.

                                Really nice feature!

                                Comment

                                Working...
                                X