The Heart of the Matter: Voice Activation
Click on this to toggle to the Brainy Read
|
The Heart of the Matter: Voice Activation |
Click on this to toggle to the Brainy Read |
| Note: theExact Word®
seeks partners in this venture. The baseline is completed; products are not. What Good Can Voice-Activation Technology Add to the Industry? Current voice-activation technology has begun to enjoy wide popularity. Voice recognition systems can "learn" differing sounds in varying dialects and from one person to another. Dictionaries for voice recognition have expanded. English, however, poses a problem: English words all have from two to five meanings apiece. And,dictionaries in the data bases build from word definitions. In this way, the data-bases "build knowledge" in an apparent human fashion: one word at a time. On the other hand, children learn something more subtle than vocabulary meaning. Children increase vocabulary year by year, but, at the same time, a child needs to know "where to put" a word. In other words, the language pattern learned by a child includes context. Adults will automatically correct mistakes that children make in word order. In a similar way, adding context to current technology would complete the "humanness" of computer voice systems. Why Don’t Current Systems Contain Context? The research initiatives have primarily sorted vocabulary meaning. For example, if I spoke two sentences using the same word in different contexts, you would "just know" the meanings instantly. a. Hand me the cup, please. b. My hand feels frozen. In order for a computer to "just know" the difference, to know which meaning to employ, to identify part-of-speech, the current data bases list all the characteristics for "hand" with each meaning. For many years, language research labs have used computer programs to "line up" words next to each other to try to determine how many times which words usually "go together" in natural human speech. Both approaches, the characteristics-list approach and the "line-up" approach, have proved unwieldy. Nor do the numbers of times words pair up provide much insight into which characteristics of different definitions for the same word would apply to which usage. How does theExact Word's® Technology Differ? theExact Word's® system uses a different premise. In natural human speech, context does not require the word and characteristic differences first. Contrarily, the definitions of words in English actually occur after word order which determines which definition applies, not the other way around. theExact Word's® has no definitions at all in the data-base, yet the program identifies correctly each word’s part-of-speech, depending upon usage. What Advantages Does a New Premise Offer? Using context would reduce unwieldy data-base searches because the need to contrast opposing lists of word characteristics would end. Voice-recognition software could easily key to the very small number of context clues necessary in English. theExact Word,® furthermore, does not require definitions. Nor did we arbitrarily choose lists of words which we say imply context. The technology rests upon a set of formulas, operating as an architecture, which have no exceptions and which pass the "test" for repeatable operations in differing environments. In other words, we organized the language structure; we did not decide upon and impose a theoretical structure. Just as humans use, repeat, and hear the same context clues and triggers in the same way in varying circumstances, so does theExact Word's® technology always recognize tricky differences, for example: "to," "two," and "too." When you hear those three words used in context, you "just know" which is which. So does theExact Word's® software, but without a data-base list of most commonly used word combinations. Another efficient characteristic of this new approach: less space required. Dreams! Whatever your wildest dreams about natural-language voice technology, we have, at theExact Word,® probably entertained them too! We hope so. We are ready to make some of those dreams into reality. With this patented technology, we think the only limits are the imagination.
Back to Voice Recognition Intro Horizontal "Heart" View
|