I bought this speech recognition kit a while ago (well, around 5 years ago). Initially I had all sorts of ideas on how to use it until I found it was a little limited. The board was made by Sensory Inc. and it was called the Voice Direct II. I say was, because it is no longer available.
I discovered after reading the sparse documentation many times over that it has 15 keywords that can be programmed in. However, for it to listen for a word in a sentence, you had to recreate the room ambiance almost exactly every time for it to recognise the speech efficiently. I also discovered that it was the very small brother to a much more expensive, better system that could recognise anything and be programmed to react in many ways.
This meant that most of my plans were pretty well shot. I had been thinking about 1812 the DRD and how it would be cool if you could whistle the first few bars of the 1812 Overture and have it reply back.
The Voice Direct module was basically a development kit board, and so after testing that the functionality I wanted was available, I decided to build my own PCB that could do what I needed. Here is the photo of it.
It is basically the development board with a transistor to switch 5v input to an Arduino.
The switches allow me to program it on the fly, however in the mode I am using, I need to press a button to tell it to listen for a word and then as soon as it recognizes the word, it immediately outputs a logic low for 2 seconds.
I can now whistle the first few bars of the 1812 Overture and it will respond. If I continue whistling it will respond with the next few bars. This will be mounted in my 1812 very soon. The Arduino sort of takes care of everything else. More about that in my next post.