[MeeGo-dev] Speech Recognition on MeeGo
jimmy.huang at linux.intel.com
Thu Dec 23 14:25:07 PST 2010
On Thu, 2010-12-23 at 16:05 -0600, Nishanth Menon wrote:
> Jimmy Huang had written, on 12/23/2010 03:44 PM, the following:
> > On Thu, 2010-12-23 at 14:56 +0800, whb_1124 wrote:
> >> Dear all,
> >> Does any one ever use PocketSphinx & SphinxBase on Meego? It is an
> >> opensource engine of Speech Recognition developed by CMU.
> >> it stand on http://repo.meego.com/MeeGo/releases/1.1/ivi/repos/source/
> >> but I can not find any upstream application of using them.
> > A sample application usage of speech recognition is the ivihome taskbar
> > in the IVI repo. Pocketsphinx installs a gstreamer plugin that allows
> > to the speech engine to connect to the mic input. ivihome connects to
> > the plugin which has a callback that returns a word if it recognizes it
> > using the defined dictionary and finite state grammar. ivihome creates
> > its own finite state grammar and diciontionary files to feed into
> > pocketsphinx. You can look at the ivihome source code to see how it
> > starts/stops the pocketsphinx and connects to the gstreamer plugin.
> > Currently, I think it is the only application that uses speech
> > recognition and resides in IVI only. Hope this helps.
> dumb question - so in theory, could we use this in a gstlaunch pipeline?
> if yes, neat!!
I think so, since the plugin is exposed as a sink I think.
More information about the MeeGo-dev