Posts: 3,415
Threads: 158
Joined: Jan 2006
Yeah, I have noticed Google Now doing this before I attached autovoice to it. The radio in the car or TV can activate it. I think part of the problem was I trained it on "Ok Google" instead of the standard "Ok Google Now". I am guessing if you use a non-standard phrase the recognition is less accurate.
I am trying to figure out how to add confirmation, but it is complicated to do in tasker. It's going to require some convoluted coding. I really wish tasker had a way to program it via a real editor vs just the phone interface. I did download the XML config file but it is very cryptic.
Posts: 2,339
Threads: 104
Joined: Apr 2009
12-24-2013, 09:55 AM
(This post was last modified: 12-24-2013, 09:59 AM by batwater.)
wuench Wrote:I am trying to figure out how to add confirmation, but it is complicated to do in tasker. It's going to require some convoluted coding. I really wish tasker had a way to program it via a real editor vs just the phone interface. I did download the XML config file but it is very cryptic.
Tasker programming is tedious that is for certain, I'm looking for a way to "RDP" into my phone or tablet and do it from a computer keyboard, mouse & monitor. There are solutions but I'm reluctant to root my current devices as I just don't have the time to mess with it.
For the confirmation consider this (presumes you are using AutoVoice); Set up your command key phrase to toggle a confirmation dialog. After "Yes / No" or whatever then set your last command ID to something, I use "Hal", and then re-trigger voice recognition. Make sure that any subsequent recognition tasks have the last command ID set to whatever you set in the confirmation task. Consider it like setting a conditional flag, only process if the flag is set...
With this method you can set up multiple "task paths" to follow based on what last command ID is set to.
-Ben
Posts: 2,339
Threads: 104
Joined: Apr 2009
Correction. If for example you were using Hal as the command ID you would set the command ID of the initial AutoVoice recognition step where you wanted a match e.g. a Yes response and then for any subsequent iterations you would be looking for last command ID to = Hal for any commands that you wanted to process after receiving the "Yes" confirmation. Clear as mud, no?
-Ben
Posts: 289
Threads: 44
Joined: Oct 2006
Here is a general question about CQC programming:
I have written many actions which are triggered by HTTPtrigger, and the same actions which are triggered by USBUIRT. How could I reorganize so I only have the actions in one place, and they could be triggered by either type of trigger?
My Other web server is Dropbox.
Posts: 2,339
Threads: 104
Joined: Apr 2009
You could write a macro that contained all of your processing and then call the macro from either the HTTP or the USBUIRT triggers passing the parms from the trigger to the macro.
Posts: 40,483
Threads: 491
Joined: Aug 2002
You can make them Global Actions. Then the actual actions just invoke the global actions, possibly passing in parameters from the incoming trigger.
Dean Roddey
Explorans limites defectum
Posts: 2,339
Threads: 104
Joined: Apr 2009
wuench Wrote:Yeah, I have noticed Google Now doing this before I attached autovoice to it. The radio in the car or TV can activate it. I think part of the problem was I trained it on "Ok Google" instead of the standard "Ok Google Now". I am guessing if you use a non-standard phrase the recognition is less accurate.
I am trying to figure out how to add confirmation, but it is complicated to do in tasker. It's going to require some convoluted coding. I really wish tasker had a way to program it via a real editor vs just the phone interface. I did download the XML config file but it is very cryptic.
Found an RDP solution, VMLite App Controller that allows you to set up your connection without root. It's only good until the phone is rebooted and then you have to hook it up to USB cable and run a
desktop app to restart the server. Serves the purpose though, keyboard and mouse on the phone. Look for
VMLite VNC Server in google play.
-Ben
Posts: 289
Threads: 44
Joined: Oct 2006
Dean Roddey Wrote:You can make them Global Actions. Then the actual actions just invoke the global actions, possibly passing in parameters from the incoming trigger.
That sounds like my solution. I gather Global Actions are in separate files. Is there any speed penalty?
My Other web server is Dropbox.
Posts: 40,483
Threads: 491
Joined: Aug 2002
It's obviously a little bit slower, but shouldn't be noticeable. Most actions are pretty small.
Dean Roddey
Explorans limites defectum
Posts: 285
Threads: 25
Joined: Jul 2009
I've been messing around with this, on 4.5.5 and for some reason my action parameter 2 on .../Test?1=livingroomlights&2=100 is "livingroomlights 100" instead of just "100". Same for "...2=0". I am using POST, and I tried GET with the same results.
Should I just remove the "livingroomlights" part manually?
