EECS 373 Winter 2007 ProjectvoiceBot
Ultrasonic sensor This sensor was used to keep track of a persons arrival or departure from the lab station. This was accomplished by watching the input data and looking for something that steadily came in from what the sensor considered to be infinity and triggered two flags that we had configured. LCD display The character display was used as the primary means of communication with the user. On the home screen we displayed a welcome, invited the user to try out our voice recognition, and finally displayed a bar graph of our ultrasonic sensor data. When someone arrived or departed, or when a spoken word was detected we would change the display to communicate that information. Audio Filtering and Conditioning In order to do voice recognition we used a technique from the middle of 20th century. Experimentation has shown that spoken words typically have three dominant frequency ranges where most of the voice data is transmitted. Furthermore, in the case of simple words you only really need to look at two of those frequency ranges. By counting the zero crossings in these two different ranges we are able to distinguish some words from other words. Parallel Port Bus In order to do text to speech we implemented a four bit data bus from the I/O expansion board to a laptop computer. With the addition of some bus monitoring code on the laptop we are able to issue different TTS commands based on the bus status. Created by Terry Dolan & Gary Wong |