Results of the design:
In our final product, we have implemented a light for the user to see the time in the dark. However due to the angle of luminance of the LEDs, it might not be able to light the display fully, hence compromising its function and performance.
The clock is extremely accurate; we have run a real-time test by setting our clock to the time of a computer clock, and leaving it on for several hours. At the end of the trail, there is no lag between our clock and the computer clock.
The usability of the voice recognition hardware proves to be a little complicated. Although, it detects the Recorded Sound under some noisy conditions, it does not work all the time as we have run several trails of test on it. When we were testing the voice recognition unit in the lab, with significant and varying amount of background noise, the unit fails to work almost 85% of the time. However, if we expose the unit just to the noise of the alarm beep, it fails to work only 50% of the time. Now, if we reduce the background noise dramatically, the unit is able to detect voices much more accurately. Due to this problem with the voice recognition unit, we are afraid that the usability of the system might not be very effective.
Another problem we have encountered is the fact that operating modes that are not 100% suitable to our needs. The first mode is recognition of a word after a press of a button. The second mode is the recognition of a word after a keyword is detected. We chose to use the first mode of operation because we would rather speak only one word to the unit and have it detect. The press of a button is activated every time the hardware goes into an interrupt. However, this poses to be a problem because the voice recognition unit would only work under a certain time frame hence not 100% concurrent with the rest of the program. The way we work around the problem is to have a LED which comes on and indicate to the user to speak only when the LED is lit.