Cambridge English Language Teaching View basketHelp
  Home > ELT > Infotech > Study online > Speech recognition
Infotech homepage
About the course
Meet the author
Study online
Weblinks

Speech-recognition

For use in class use the printout version of this page.


A. Read the article and then complete the exercises below.

Speech-recognition

Mobile phones and the Internet have changed the way we communicate. However, we still need to use the keyboard and the mouse to communicate with computers. When shall we be able to interact with PCs by voice? Speech recognition systems can make this possible soon; in fact, speech companies claim that their programs let you dictate, control your PC by voice command, and convert text into speech.

The most popular voice recognition packages are IBM ViaVoice, Dragon NaturallySpeaking, and L&H Voice Xpress. If you decide to buy one speech program, look for these features:

  1. All the programs let you dictate text directly onto your word processor, spreadsheet or e-mail. They claim an accuracy rate of around 98 percent. The technology is particularly useful to dictate notes, business memos, letters and e-mail.

  2. Speech programs are usually more accurate if you train the software by reading aloud for about 20 minutes. This process involves reading sample text and teaching the program to recognize words that are not in its built-in dictionary (e.g. proper names, acronyms, unusual words). Most products let multiple users personalise their vocabulary and pronunciation.

  3. They offer useful voice commands for editing and formatting (e.g. setting font size and style, text alignment etc.). However, it’s a bit difficult to make corrections by voice; using the keyboard may be faster and more accurate.

  4. Speech software allows you to use commands to launch programs or to navigate around windows and menus. You can also tell your programs to open a file, save it in a particular format or print it.

  5. Some systems let you search the Web by voice. The web version of IBM ViaVoice enables chat room users to chat using their voices rather than keyboards. Dragon NaturallySpeaking lets you browse the Web by speaking URLs and links, and dictate into online forms and fields in Microsoft Internet Explorer.

Speech recognition requires a powerful processor (500 MHz Pentium recommended), 64 MB of RAM, a good sound card, and a high-quality headset microphone.

Speech recognition technology has improved a lot and its potential is enormous, but some experts consider that it is not ready for common use yet. Some day, however, we’ll be talking to our PC naturally, like a friend.

Speech recognition program from Dragon Systems, Inc. Speech recognition program from Dragon Systems, Inc.

A. Tick the features mentioned in the article.

Activate menus by saying the words on the screen
Execute applications by voice
Manage databases
Configure multiple voice profiles
Read documents to you using a text-to-speech system
Surf the web by speaking
Create and compile a computer program
Take dictation with accuracy
Design graphics
Accept spoken commands

B. Language work - Modal verbs. With the help of the text, fill in the gaps in these sentences. Use the correct modal verb from the list.

can
(ability)

must (necessity)

should (advice)

could (possibility)

will (prediction)

would
(in hypothetical conditions)

  1. With many voice-recognition programs, the user first train the software to recognise individual pronunciations.

  2. You dictate into the program’s text screen, or dictate directly into any Windows application.

  3. If you intend to do a lot of dictation, you get a high-quality headset microphone.

  4. Speech-recognition software help children with special educational needs.

  5. In a few years' time, everyone use voice commands to interact with computers.

  6. If I could afford it, I buy a new PC and a speech-recognition package.