Training VC to recognize a command from a 3 year old?

[alk23k]

How I am using VC:

I setup a button on my 9 button Keypad Linc (Smarthome Insteon Lightswitch Scene Controller) to control Voxcommando.  I set it up in my kitchen so I can tap it and VC goes in listen mode and the LED lights up on the light switch to let me know it's listening.  This way I have a visual indication when my AV Receiver is off.  Of course I can also turn the switch off or let it time out and VC goes to ignore and LED goes off.  Pretty handy

My next project is to set it up so that when my 3 year old walks down stairs in the morning:
The Main Room camera detects motion > Blue Iris (Cam Software) sends event to VC > VC turns on TV etc and welcomes my 3 year old and asks him what he wants to watch.  He can then start his favorite show, Transformers of course. 

Question on this: any way to train a single Action in VC to recognize my toddlers voice?  There is no way he could go through a whole windows speech training session on his own.

[jitterjames]

That's cool.  You should post a video to show off your fancy button / light setup!

Rather than training you might just want to work on choosing a phrase that works for your son.  Generally a trained profile is only really useful for one person so unless you want to have a special profile just for him the training is likely to cause problems for you in general.  It is also important that he understands how to talk to a computer (something many new VC users of all ages have trouble with!).  You need a silence before your command, and then you need to say the correct phrase clearly.  Consider it an opportunity to get him started on good diction early in life.

That said, if you right-click on a command in the tree editor you can train with that 1 command by selecting "Train speech profile" from the context menu, or you can select a command and press Ctrl-T on your keyboard.  You should start with the rejection level quite high and lower it until accepts the command.  There is also an action for switching between speech profiles so it could switch to his profile when he triggers the motion sensor in the morning.

I don't actually know how well it will work for you but you can give it a go and let us know!

[jitterjames]

There is no way he could go through a whole windows speech training session on his own.

You might not be able to get through the whole thing, but it might be worth trying.  I'm not sure how it works but it might still have an effect even if you only do the first few sentences.

[vulcanjedi]

This is pretty awesome and I too want to see it.
My Vox/windows had a tough time w/ my nephews and nieces  when I brought it all home over a few holidays.
Cheers, I hope your son is watching the original Gen1 Transformers!!

The only thing I could think of in terms of your 'single action' is to spin up a 2nd instance of VC? w/ just 1 command loaded?
Maybe set your Blue Iris command to terminate the main VC and spawn your sons in the am when he wakes up and triggers the motion alarm?

[jitterjames]

You could also enable / disable groups, but the problem with having only one command is that the false positive rate will rise significantly.

[alk23k]

That makes sense.

I do want to have a few actions available to him.  I am going to set it up so that it suggests 3 or 4 shows to him and then he can respond with one of the show names.  He is a pretty smart kid hopefully we can figure it out together.

I will have the command trigger an autohotkey script that I have written which will execute the most recently recorded version of the show he is interested in.

jitterjames per your suggestion I think I will have it also temporally disable non relevant groups.

I do wish there was some way to do sound recognition and not just voice recognition.  That would be a supper cool feature.  It would be cool to record commands and have it recognized based on the recording.  If I had this I would also set it up so that when my home phone rang VC would take action.  Or if the doorbell rang pull up the front door cam etc.  Sound recognition would be cool.  Just thinking out loud.