|New Mexico Supercomputing Challenge|
Challenge Team Interim Report
Voice-Activated Flat Television Set (VAFTS)
At the beginning of the school year, we came up with the idea of planning a voice-activated flat television set for our project because we were thinking of the disabled people and the safety of our environment. To do this we had to research the components of voice recognition and record the three languages we plan to put in for people who speak English, Navajo, and Spanish. We each did our part of research on the project. One group member will be doing the programming of C++ and HTML formats. Two members worked on finding and recording the three languages to be used. Another member is building a model or drawing of our work. All the members did their parts along with helping out with all the research of voice recognition and brought it all together for our report.
First we had to find out how the activation works. There are two general classes called "Speaker-dependent" and "Speaker-independent" recognition. For Speaker-dependent, the speaker trains the system to recognize his or her voice by speaking each of the words in the inventory several times. For Speaker-independent, the device is not trained by the speakers’ voice, since it was previously trained on samples from many different speakers.We found lots of software about voice recognition but we still don’t know how to put it in a television chip. There’s software called "Smart Speak." It’s easy to use for programming in Visual Basic and C++ programming. We are thinking of using this software to help us out with our programming part of the project. We also found a program called "Phonemes," which explains how voice recognition works. This software compares the person’s voice with samples of thousands of other voices. Matches may still be questionable from the words, in which "Phonemes" looks for common combinations of words that make up a word. For example, "Phonemes" may look for the word ‘dog’, it will look for little combinations in the word such as duh, aw, guh. This makes up the word dog. Once the software identifies the word combinations, it may still find problems because of homonyms. The software then compares the homonyms with the rules of sentence structure. For example, they may use the word ‘their’ and at the beginning of the sentence, the word ‘they’re’ may come up.
Mr. Barak A. Pearlmutter volunteered to be our mentor for our project. He is an assistant professor at UNM and will be helping us on our project of Voice Activated Flat Television Sets. His first organization is Computer Science, his second organization is at the Department of Neurosciences.
For questions about the Supercomputing Challenge, a 501(c)3 organization, contact us at: consult1516 @ supercomputingchallenge.org
New Mexico Supercomputing Challenge, Inc.
80 Cascabel Street
Los Alamos, New Mexico 87544