11 Aug 2005 (updated 11 Aug 2005 at 02:39 UTC)
»
Going in Deeper...
If any of you were wondering why I wasn't updating, it was
because I was on vacation. But now I am back, so on with the
updates.
Recently I have decided do go deeper with my voice
recognition journey. Instead of using the Microsoft speech
SDK, I am going to be developing my own Speech Recognition
engine. My reason for doing this is although the MS Speech
SDK is nice, it doesn't provide the accuracy that I am
looking for.
I know, this seems as if it would be impossible, but I have
worked it down to a science. The way I will attempt to do
this is I will have my Microphone input stream inputting to
an array of numbers, then I will compare that array to
another array for specific commands.
If you are asking how this will be more accurate, then I
shall explain. The original arrays that I will be comparing
against is the specific words, spoken by me, so the input
and the comparing arrays for specific words will be
extremely close already, also I won't have millions of other
words that the engine is trying to match the input against.
Only the ones that I setup.
Right now I am leaning towards using the DirectX Voice input
libraries as my voice command input stream. By if I can find
something more efficient, then I go for that.