speech to tweet

I was at #mcn2 (media center nottingham) on Saturday.
http://sites.google.com/site/mediacampnottingham/

Glenn Le Santo (@lesanto) gave a presentation about the last 10 years and the future 10 years of the internet.
1 thing he mentioned was how he would like to be able to tweet whilst driving, essentially convert speech to text then tweet that text.

I've been bored, so decided to have a play with this idea. I am basically putting together open source libraries to create this.

The speech recognition library being used it call PocketSphinx (http://cmusphinx.sourceforge.net/), using GTK and GStreamer it was possible to create a simple Python GUI which displays the text of what you're speaking (well sort of).

I then use the python-twitter (http://code.google.com/p/python-twitter/) API to send the message generated to twitter.

This was just something to do, the actual speech recognition is well off. this may be due to a shitty microphone? PocketSphinx is something i've been meaning to play with for a while now, i just haven't every gotten round to it. This was my first experimenting with it, from reading there are some training (sphinxtrain) which you can conduct to get better speech recognition. Its a nice idea though, would be nicer if it worked better :)

I will post the python code and details of how to get this working, I will post a link when it is on my website.

Category:

Science & Technology

Tags:

• #mcn2
• pocketsphinx
• python-twitter
• speech to text
• daveisanidiot

[AZTaitjcize]

Hi, have you get this working? or can you provide the details to the usage sphinx library? Thanks!

[d4v3]

@AZTaitjcize i didn't really do anything other than this. the sphinx library which i used was a python library, in ubuntu it was just a case of searching sphinx and it showed up in the package manager. Then included that library in the python script.

[AZTaitjcize]

@d4v3 Thanks for the tip :)

I didn't notice that it was already in the ubuntu repository