Early Android SDK brings voice recognition

Early Android SDK brings voice recognition

If I were to name a feature Linux lacks compared to MacOSX and Vista, that would be voice recognition. As far as I can tell, there is almost no working voice recognition software for Linux.

Here comes Android, the 1.5 early-look version of their SDK brings official support for voice recognition. That’s something Linux distribution may really want to grab and re-use.

Update: Four Feet‘s Chris Hager has been kind enough to point out that the speech recognition in android is based on the proprietary technology of Nuance. Nuance technology is well known to be effective and real-world proven (I even heard about US courts using it to transcribe their recording of the trials). Unfortunately, I see no chance of being able to have it on Linux. Not for free at least.

Update2: this article claims that the speech has to be sent over the internet to be converted to text. That shed a new light on the matter, and contrasts with Chris opinion.

Android voice recognition

As highlighted by this post, unofficial support was already available in Android SDK 1.1., with code being as easy as:

Intent intent = new Intent("android.speech.action.RECOGNIZE_SPEECH");
startActivityForResult(intent, 0);


	protected void onActivityResult(int requestCode, int resultCode, Intent data) {
		...code for checking resultCode and data intent...
	ArrayList results = data.getExtras().getStringArrayList("results");

You could call your pc by name (let’s say called it Kubrick) and:

– Lying on the sofa and watching a movie. Suddenly, the phone rings
Kubrick STOOOP !
computer: (the stop multimedia key action is triggered, the movie stops)
me (after carefully explaining that yes, I really liked her new shoes, really):
Kubrick PLAY !
computer: (play  multimedia key action  triggered, the movie resumes)
Everything without having to stand up to to reach the computer.
– five minutes later, the phone rings again (oh shit !) … to be continued ;-)

Another use case may be the netbooks, where the keyboard and trackpad are so costrained that may be more easy to launch programs by name than digging the menus or pressing hotkeys.

Comments welcome. :-)

This seems like a great time to subscribe my RSS !

10 responses to “Early Android SDK brings voice recognition”

  1. Mike

    Another useful thing would be writing text with voice recognition. But that needs a really good recognition and a large dictionary.

    The netbook thingy you mentioned would look awesome in the subway. :D

    >As far as I can tell, there is almost no working voice recognition software for Linux.
    Still alpha but looks promising: http://www.simon-listens.org/index.php?id=122&L=1

  2. Stefano Forenza

    Thank you Mike, I’ll check it out. The domain name rocks :-)

  3. Chris Hager

    More details about the speech recognition API, including the documentation and a working example can be found here: http://www.4feets.com/2009/04/speech-recognition-in-android-sdk-15/

  4. Stefano Forenza

    @Chris: seems nice a nice article. I have a question for you: I know some part of Android are still – unfortunately – proprietary. Are the speech recognition libraries open source ?

  5. Chris Hager

    @Stefano: thanks for your reply and question! No, the speech recognition engine is not open source, but proprietary software. It’s just a third-party activity implementing the technology of Nuance, pre-installed on the phones in binary form.

  6. Stefano Forenza

    Nuance, so DragonNaturallySpeaking. A pretty good choice. Too bad it’s proprietary :-( .
    Thank you!

  7. Stefano Forenza

    This article [ http://www.androidandme.com/2009/05/news/cupcake-voice-recognition-now-available-for-testing/ ] claims the API just calls a web service. If anybody has any information on it, including an eventual license please post a comment.

  8. Robin

    As soon as enough people visit VoxForge.org to donate their speech we would no longer need to be envious towards closed source speech recognition (SR) solutions. The lack of open source licensed speech is the major obstacle to developers of OS SR solutions. Let’s change that!

  9. abdelatif

    salut , je suis entrain de realiser une application pour programmer un clavier vocale (au lieu de tapper une lettre par exemple “A”, l’utilisateur va la dicter directement puis l’application doit reconnaitre la lettre ” A” puis elle sera afficher a l’ecran du smartphone ) = pour faciliter la saisie des sms; j’aimerai bien que je trouve un aide sur les api de reconnaissance et synthese voclae pour android.
    la deuxieme chose est : est ce que l’emulateur d’android prend en charge les applications de reconnaissance et synthese vocale;

  10. Karim DURAN

    Hi Stefano,

    I think you’re wrong. There is a very good speech recognition engine on Linux. Its name is “SPHINX”. It’s open source and comes from US Carnegie Mellon University. You can download it from sourceforge, compile it, and download language models and dictionnary.

    You can also install it (binary release) on your system. Go to synaptic -> search -> sphinx -> install ( bin,library, doc, dev).
    You have exemple programs to test it.
    You also have sphinx4, a full java version; and sphinxpocket for mobile devices. An android demo is also provided.

    Try it.


Leave a Reply


Don’t forget to Subscribe

Latest Activity


  • Facebook does man-in-the-middle attack
    It’s never too soon to delete your Facebook account. And never a bad idea. I deleted mine some month ago and didn’t regret it once. According Read More
  • LOL
    https://lists.ubuntu.com/archives/ubuntu-desktop/2011-November/003467.html many lessons to be learned there * Share it !
  • Deja vu
    Reminds me of something. Share it !
  • Google this is ridiculous
    Google you’re doing it wrong. Very wrong. This is utterly ridiculous. It’s a screenshot with a standard Firefox browser, in the standard screen resolution (1280×800), on Read More
  • A step back from the open source
    One month ago, I created my first Android app. While the app was a paid one, the reception has been outstanding. I’ve gotten a fair amount Read More