Voice Recognition

If typing is not one of your strong points or if you need your hands free for things other than typing, voice recognition may be ideal for you. People who have repetitive strain injuries or disabilities that prevent them from typing well or typing for long periods of time may also find voice recognition advantageous.

To decide what type of voice recognition you want you must first decide what you want voice recognition to do. The simplest level is Command and Control. These programs allow you to operate windows or other operating system using voice navigation commands and voice macros. Word processing and similar typing is specifically for online use with browser, email, chat, and messaging software. At the other extreme are full-featured dictation systems that allow you to dictate into word processing or spreadsheet programs, write email, and participate in chats using voice recognition. Most of these will also include Command and Control features.

If you will be using voice recognition in a field with very specialized vocabulary, there may be versions designed for your profession. Specialized versions are available for medicine, legal, and emergency professions. Often these versions also include soft

. . .

View the whole essay, without interruption

Check the requirements for each version to make sure it will work on your existing hardware and software and that it meets your basic voice recognition needs. There is a mobile version that includes a digital recorder as well as a version designed specifically for teens. A Linux SDK with Java code is also available. Pattern matching algorithms are used to match the captured speech with the words in the stored templates or models. Voice Xpress is also available in five languages plus US English and specialized versions in Cantonese and Mandarin.

Dragon Systems has the Dragon Naturally Speaking series of products for Windows 95/98 and Windows NT. Adaptive systems will work to some extent with most speakers but work best with additional training such as reading text aloud. Their software includes the ConText tuner.

Voice recognition identifies patterns in the acoustic speech signal and associates them with text. Most people prefer continuous rather than discrete. Dragon Naturally Speaking has a free SDK for SAPI or ActiveX controls. Most of the current commercial voice recognition products use continuous voice recognition.

Approximate Word count = 1066
Approximate Pages = 4 (250 words per page double spaced)

Simply subscribe to view this paper, and 100,000 others.