What you would do differently next time:
If more digital signal processing experience was acquired, we would definitely make our own voice recognition device, by buying only the voice recognition chip, and not a kit, as we did in this project. If we build the whole voice recognition unit from ground up, we would be able to make the unit be more lenient to noise and the voices of the different users. Currently, the unit we are using does not seem to have a very great margin of error with different sounds, although it is the same word. It is doing a too good job in identifying the word spoken out by the user. For example, if we say “Stop”, in an elongated and lazy way and stored in VoiceDirect 364TM, and we input saying “Stop” quickly and short, it will not be able to detect the input and outputs an error for the user.
As with the modes of operations, may be next time round, we would try the second mode of operation. That is, using a keyword to activate the voice recognition, and compare the difference in usability as compared to the mode of operation we chose now.
We would also seek other opportunities in finding a bigger LCD with a backlight, mainly for the reason that it would save us wiring procedures when packaging the clock unit. The LED method in lighting the LCD is a good alternative, but it does not serve our purpose as much as we had hoped for.