Google and its offline voice recognition engine.

Google announced these days, at his Google IO event, that the upcoming Android 4.1 “Jelly Bean” will come with an offline voice recognition engine. Who tested this offline mode, seems to report that it works really good also when used to compose messages, and that it seems to work also for non-english languages! This means that, soon, Voice Control user’s will be finally able to use it completely in offline mode without any restriction compared to the online mode!

This obviously makes me think about the Voice Control embedded offline voice recognition, that requested a lot of work to be functional, and that still needs a lot of work in order to be enough usable, because currently I’m receiving really bad feed-backs from beta testers about this feature, without considering that it was planned to work only with a limited set of commands.

So, considering that Google has developed an offline speech recognition engine that works definitely better than the one embedded inside Voice Control, and that maybe there are some possibilities that the offline mode of the Google’s engine would be also released for non “Jelly Bean” Android versions, I’m planning to freeze and drop any further work on the embedded offline engine, removing it from the upcoming release in order to focus only on improvements of existing and new features.

I hope that this decision will help me to release the next version in less time.


5 Comments on “Google and its offline voice recognition engine.

  1. So now you plan on stealing others work. Considering you lie in your description (yes son, you must take your eyes off the road to press buttons), and then refuse to give a refund, I call you a cheat. Considering who I am, I’m going to make sure Google, etc knows about your bullshit… you’ve made an “enemy” for life…

    • Hi jeff,
      I never refused to give refunds to everyone, and every time that someone have a problem I’m the first that suggest to apply for a refund.

      You are free to request a refund sending to me your Play Store order number or using the confirmation mail send from the Play Store, and I will be happy to apply for a refund.

      If you are referring to the Voice Recognition engine, I never said that Voice Control has a proprietary Voice Recognition engine, and I already said that it uses the Android Voice Recognition features that are freely and expected to be usable by other applications like Voice Control (and others).

      I also explain in the app description that Voice Control can be started by a Bluetooth Headset that supports the Hands-free bluetooth profile, but that due compatibility problems between Android phones and Bluetooth devices that doesn’t depends on Voice Control itself, it is suggested to try the Free version of Voice Control, that can also be used as an evaluation version installing a trial license that will unlock the Paid features for a week, before to buy the Paid version.
      It is obvious that, if you don’t have any bluetooth Headset, you need to look at the screen in order to start it… However I added some alternative ways to start Voice Control in order to help users without a bluetooth Headset, like a big home widget, or the possibility to start it long-pressing the search physical button.


  2. RE: “Voice Control user’s will be finally able to use it completely in offline mode”:
    Clearly you haven’t tried many of the features of Voice Control, which depend on online access to be of use (e.g., navigation).

    RE: “works definitely better than the one embedded inside Voice Control”:
    How can you know that when you apparently haven’t been able to test it? The existing online one actually works very well and your premise that the new offline capability will beat that of online backed by much more powerful and capable machinery is simply just unfounded conjecture.

    RE: “Google has developed an offline speech recognition engine that works definitely better than the one embedded inside Voice Control”:
    – Their is not one “embedded” inside Voice Control. It’s remote. And you clearly have no real data to suggest offline works better than the prior online version. You have no idea whether it’s parred down, makes sacrifices for less computational resources, and the like.

    If you don’t have data to back up your babble, please avoid pretending to know something you don’t.

    A Google Employee

    • Hi Bilbo,
      I think that you have misunderstood this post: I was talking about a “self written” offline voice recognizer that I was developing (based on another open-source project), in order to embed it inside my Android application called “Voice Control” (available on the Play Store at this address, and that I stopped any further developing of it because Google announced that, in Android 4.1, there will be an offline voice recognition engine made by Google itself.
      Instead, it seems that you understood that I was talking about Google’s Voice Action (or Google Now?), but it is not the case.

      As an update to this post (not related to your reply): I was able to test it on my Galaxy Nexus, and it really works good for message dictations, but, sadly, it still don’t use offline recognition data when used by external applications (at least when used through the RecognizerService class).


Leave a Reply

Your email address will not be published. Required fields are marked *