Ideally, once voice recognition applications are installed and configured, the user is able to simply dictate documents, email messages, database entries, and numerical values for spreadsheets with reasonable speed and accuracy.
Nevertheless, the equipment may fall short if
- The computer is in a noisy environment and the noise-cancellation technology of the sound card and/or microphone is not of high enough quality
- The computer lacks a fast CPU (e.g., 400 MHz or more) and enough memory so processing the speech becomes extremely slow
- The user has a strong foreign accent or speech impediment, and even multiple registrations cannot completely compensate for this
- The user wants to enter foreign words and phrases, technical and scientific terms, or other speech that is not easily recognized by the software
- The user needs to pause mid-sentence, perhaps to catch his/her breath or read from a manuscript
Depending on the type of speech, the user may have to spell out many letters (possibly requiring use of military alphabet words such as alpha, bravo, and charlie) in order to achieve the desired level of accuracy.
For the Sight Impaired
Visual impairments are perhaps the most difficult to accommodate as the computing environment is highly graphical. Screen magnifiers can help users with moderate visual impairment by magnifying the entire display area two- to four-fold and allowing the user to selectively enlarge portions of the screen or to create a partially magnified split-screen display or a zoomed "picture in picture." For users unable to see magnified displays, refreshable Braille screens are available along with various screen-reading (text to speech) applications. Some screen readers can vocalize both screen content (e.g., menus, Web pages, and spreadsheet cells) and user input entered via the keyboard (standard or Braille). Alternatively, a screen reader can be coupled with a voice-recognition application, so that both computer screen display and user input can be handled verbally.
Considerations for Web Development
According to an eWeek article entitled "Web Blind Spots" from 10 April 2000, between 95 and 99 percent of Web sites are still not completely accessible to the visual, hearing, and/or mobility impaired, especially sites using proprietary, customized software. The nonlinear organization of many Web pages, and the generous use of tables, frames, graphics, buttons, and icons in place of simple text make it difficult to effectively use screen readers and keyboard-based or voice-activated navigation. However, use of the "ALT" tag in a Web page's HTML code can help vendors make their Web sites accessible and attractive without breaking the bank. Supported by most browsers, the "ALT" tag allows Web programmers to assign brief text descriptions to images and navigational icons such as buttons and links. With a screen reader, a visually impaired user can get an audio description of the ALT-tagged icon and can more easily navigate the site. A Web site can also be designed to offer the user a low bandwidth version, consisting of single-column text in place of graphics, tables, frames, and other formatting enhancements.
Other techniques for creating accessible Web sites are discussed in the World Wide Web Consortium's document "Authoring Tool Accessibility Guidelines 1.0," available at www.w3.org/TR/ATAG10. In addition, Web designers can assess the degree that their sites are accessible using the "Bobby" test, developed by the Center for Applied Special Technology (www.cast.org).