Wednesday, March 28, 2012

First post from the android phone

This is an experiment. I have enough trouble getting the keys on a full size keyboard. I have no hope it all exiting the keys on this tiny little keyboard. F*** away android has a voice recognition system that simply translates my voice into words. To show how well it works I am not gonna correct anything on this blog and you can just see what mistakes have been made


  1. OK I am at a full size keyboard now, and I just have to comment I never said "F*** away". I said "Fortunately". Also, I have trouble "hitting" the keys, not exiting the keys. But otherwise, very high level of accuracy I think, and a great time saver, maybe even faster than typing.

  2. You write, 'maybe even faster than typing' ... Well, that depends on how fast you normally type ... LOL!

    The only time I've resorted to voice recognition was during a project to put a school yearbook online. The yearbook was printed on a 'pebble' finish paper which presented scanning difficulties, and the number of proper nouns, especially peoples' names, posed challenges for the OCR software I was using.

    But I was surprised at how well Dragon's NaturallySpeaking did with all those lists of class members - and that for a school with high 'ethnic' content. My first experience with voice recognition, and I was quite impressed.

    I can only assume that, in the intervening decade, this class of software has improved significantly, including efficiencies for use on the small processors in smart phones.

    So ... why does 'voice dialing' on my cell phone never work right? ... LOL!!

    1. The voice recognition is much faster than me typing on the phone's virtual keyboard. I just found out that this is not being done on the phone's processor, which I think is 1 gigahetz speed. Instead, it is being done by Google's processing power through the internet connection.

      BTW, I was not the one who put ***, I think that was Google's software.

  3. Interesting. I had not kept abreast of this VR stuff but, indeed you're right ... for Android, the audio is shipped to a Google server for text conversion ...

    'responds to a RecognizerIntent by displaying the "Speak now" dialog and streaming audio to Google's servers -- the same servers used when a user taps the microphone button on the search widget or the voice-enabled keyboard.' ...

    This is as fascinating to me as when I discovered that Google Translate is not rule-based but simply drives off statistical analysis techniques ...

    I suppose I'm simply beginning to show my age (and my 'stone age' data processing preconceptions ;-)