Speech recognition Maemo?
 

Hello,

I am wondering if anyone knows of any API's in any language which provide speech recognition and could be applied to the N8* series? I have been looking for something in Python but didn't come across anything.

Does anyone know any existing apps that use this functionality or if it would even work on the IT?

I am thinking about making tools that let you record macros which can be executed via verbal commands.

Cheers.

Re: Speech recognition Maemo?
 

Speech recognition in linux is rare to begin with.

You could look at http://freespeech.sourceforge.net/ or http://en.wikipedia.org/wiki/Julius_%28software%29.

Re: Speech recognition Maemo?
 

This might be your best bet to port:
http://pocketsphinx.org

Note the section on "How do I make it fast?" Even once ported it may require some fiddling to get it to run in real-time.

Of the Linux-based speech recognizers I have mainly used Sphinx4, which is Java-based and probably memory-prohibitive for the NITs.

I don't know what Julius's memory/processing requirements are, but I think it is worth a look too.

Might be worth poking around
http://voxforge.org
too.

Re: Speech recognition Maemo?
 

@jkoder

Sounds like it might be possible. PM or email me if you want a little help - I'm not really a programmer but I have dabbled in the good old days...

Clive

Re: Speech recognition Maemo?
 

Your best bet might be to look at KDE that has speech recognition abilities. Porting one of the above to KDE might be easier than directly to NIT. I use flite to have articles, websites, doc, etc... read to me by a Stephan Hawkins like voice in KDE. It's pretty cool. Dan

Re: Speech recognition Maemo?
 

@ Dan

I think jkoder was looking for speech to text. Flite is text to speech though I would love that also so I don't need to carry my reading glasses with me all the time!

C

Re: Speech recognition Maemo?
 

flite is already ported. Maemo mapper uses it to read directions out loud.

Re: Speech recognition Maemo?
 

Just wondering... my Nokia 6110 Navigator works quite well with speech recognition: there's a well-defined set of commands, and its very easy to use it.

Instead of reaching for the stars and make the tablets use speech recognition for text input: wouldn't it be much easier to use it instead of menus and clicks (long click included...)? Would be cool to pair my bluetooth headset and say something like "browser. internettablettalk. full screen. scroll down. menu. send to. e-mail. ..."

This could end the "ginger-friendly" vs. "optimized for stylus" once and for all :D

Re: Speech recognition Maemo?
 

What about sphinxbase and pocketsphinx?