New Fork: Voice Control For XBMC Remote

tested the following.

"play movie Hitman" : works flawless
"play movie Contagion" : works flawless
"play movie Harry Potter and the Chamber of Secrets" worked on the second atempt.
"play movie Harry Potter and the Deathly Hallows Part 1" voice recognized "play movie Harry Potter and the Deathly Hallows Part 1" , won't play the movie though. ( Movie Title: Harry Potter and the Deathly Hallows : Part 1") Maybe some issues with the ":"? ( same issue with X-Men: First Class)

Sweet. Did albums work for you as well?

If you are having problems with accuracy (eg needing to say stuff two times in a row etc) then you can go to google voice search and turn on personalization - so it will learn your accent and pronunciation, and hopefully get more accurate.

Titles with symbols like ":" definitely will not be working yet. I have to think about how I'm going to match when there are strange characters in the title.
I also haven't dealt with numbers very well yet either: the whole "1"/"one" thing needs a solution.

"play album american idiot" works just fine. Realy great to see someone finaly working on this =).

Hi,

it would be realy nice, if you could ad a "stay listening" or dock feature, which constantly waits for an incomming command like "XBMC pause"... this way we dont need to push a button and simply could pause with nachos in our hands

Greetings

This is very exciting! thanks for looking into this feature!

Haven't been able to get it to recognize any albums yet.. byt the xbmc remote isnt listing albums right now anyway.. possibly a nightly build issue?

Hi Limx:
I'll take a look at maybe doing a continuous mode listening feature for dock mode (it would certainly drain lots of battery if it wasn't in the dock). However, I'm not 100% sure if Google's voice recognition api allows this - their servers might kick us off if you start a connection to their server and hold it open for too long.

I do like the idea of a continuously listening feature. In a couple of years when phones get powerful enough and we can do the voice recognition on the phone instead of in the cloud, we'll have the opportunity to replace the google voice api with a port of something like pocketsphinx, which will make continuous mode easier.


Hey Skabyss: don't know much about the xbmc nightlies- I'm running xbmc version 11. But if the remote isn't even listing albums, you are probably right that there might be some issue with the nightly builds.

Been a bit busy lately, but hopefully will have a new alpha out with a couple of new features in a fortnight or so.

Thanks for the effort! I'm pretty excited about this project.. But like another user said, you should check out voxcommando too as it can accept text strings over a network and interpret them into actions.

Thanks for the suggestion, I had a look at voxcommando, unfortunately it is Windows only.

I use Linux on my htpcs. The reason I do this is because Linux comes with music player daemon (mpd) and pulseaudio. Using these two things together gives me the ability to play the same music perfectly synchronized across the network in different rooms. This is something that is an absolute killer app for me, and I have no idea how to do it as easily in Windows (if at all)

Linux also has other advantages, but that is the main one I wouldn't be willing to give up.

So a Windows only solution like voxcommando is not really an option for me.

I agree I rather have a cross platform app. I have windows, linux, ios and osx machines in my house.

This is true, I had forgotten it was a windows only app.. But I did buy the software so maybe I'll go kick sand at the dev. until he opens things up more