![]() |
Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
where to put these codes ??
pocketsphinx-utils pocketsphinx-hmm-en-hub4wsj pocketsphinx-lm-en-hub4 gstreamer0.10-pocketsphinx python-gst0.10 python-gtk2 python-gobject python-cairo espeak |
Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
Quote:
|
Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
Quote:
I replace ( * ) in *.deb with package name ? |
Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
Quote:
|
Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
dpkg -i *.deb .. couldnt open debian archive :(
|
Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
solved .. thanks a lot for this app
|
Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
??????
It doesnt understand any single word well it just understand (go away) .. =p |
Re: [DEVEL] Saera: Siri clone for Maemo 5
1 Attachment(s)
Quote:
|
Voice control for Maemo5 and Harmattan
Hi,
like taixzo I developed a little voice control app in Python. It's named Voicy, yeah not very original. Unlike taixzo's app “Saera”, which is really cool, funny and implements some AI, my app -which I have not made public- only tries to be functional. It just can launch and control other apps. Besides, I wrote it mainly to control navit by voice, which works pretty well. (Currently implemented navit commands: Navigate home: looks for a navit bookmark “Home”, sets this as target and starts navigation; Stop navigation; Zoom in/out, Pitch up/down; Layout car/bike/car-dark) My program also makes use of pocketsphinx, like “Saera”. Since the voice recognition of pocketsphinx unfortunately is not perfect, I took the following approach: The program only uses a limited set of very distinct commands at a time, but it changes from one set to another, depending on the current foreground task. Example: Ask Voicy to lauch the mediaplayer. After the mediaplayer got started, the set of available commands is not automatically(!) changed. But almost immediately after mediaplayer actually got the new foreground task, Voicy will recognize this and in turn will switch the set of available commands accordingly. This means, also when the user manually(!) launches another app or brings it to foreground respectively, the software will change the command-set. When the current foreground app is unknown or Voicy itself, the command-set of Voicy is active. Moreover I patched pocketsphinx to deliver a score value. When the patched version of pocketsphinx is used, which is optional, a command is only carried out at a minimum score level per command. The core program is not finished yet, I would like to add little more functionality. Currently there are other things with more priority, so this might take some weeks or months. It should be pretty easy to add new functionalities for arbitrary other apps, provided there is some kind of interface to control them, otherwise Xdotool is an option. The Voicy core program is like a -let me say- little framework. During initialization it looks for control-files in a certain subfolder structure and dynamically imports them. That means, adding functionality for a new app only requires to write some code for it, mainly “do this or that on a certain command”, add the required language model files for pocketsphinx and place these files in a new subfolder. Thats it. If you want to enable Voicy to launch this program, (unfortunately atm) you need to edit the command-file for launching apps and add a new entry. :confused: Well, my question is, are there still users of the N900 which are even interested in this little piece of software? (If there is anyone, perhaps just click the Thanks! - link as a feedback.) If so I might take the time to make it public. Who knows, perhaps there are even people which are interested in coding/adding new controls :rolleyes: |
Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
I am quite interested in that. I've found it hard to actually make Saera do anything useful, because it's trying to do everything at once.
|
All times are GMT. The time now is 13:50. |
vBulletin® Version 3.8.8