Recording voices
and finding the
exact word to be used
as the input signal
Jaime Tayamora
back to CE150
Recording sample voices that will be used as input for the PC-controlled car seems a very easy task at the start. Equipment that can record voice is easily available, much more, persuading an individual to make his or her voice the sample to be used is not a very difficult task at hand. It was obvious at the moment that almost all of us wanted to say the words 'go,' 'left,' 'right,' 'back' or 'stop' and see the car moved.
The recording process proceeded without much difficulty, except for certain factors, which are quite inevitable especially for a surrounding that we had in the laboratory. Since other groups were also finishing their projects and were all hurrying to see the results of their jobs, technically, the signal-to-noise ratio is very low and this affected the input for quite a few sampling.
The device used is a PC microphone, with a small plastic stand and handle, connected to the soundcard at the back of the CPU. This would later on serve as the sole connection of the controlling person to the RC (remote-controlled car). The words intended to be used to control the car are 'go,' for moving forward; 'left,' for turning left; 'right,' for turning right; 'back,' for moving backward and 'stop,' for making the RC halt. At least ten of each of these words were recorded via Windows Sound Recorder and stored as a .wav file. These voice files will then be processed further using MATLAB in order to be analyzed.
Consequently, the recording process would have ended promptly after the sample voices are saved if not for some serious problems encountered after the initial samples were analyzed. Through many various processes done in the MATLAB, it was found that not all of the original words could be used to control the RC. Only the words 'go,' 'left' and 'stop' remained and substitutes for the other command words have to be formulated.
Eventually, instead of using the words right and back, only the letter 'r' is uttered for 'right,' and the letter 'b,' pronounced with a high pitch, is used for 'back.' Another set of the new commands were recorded and then integrated in the already running program in the MATLAB.
|