Speech Recognition Software

Speech recognition Any technology that converts spoken words into text or commands. software converts what you say into text or mouse commands. It benefits people who have physical or cognitive difficulty using a keyboard to create text.

This article explains how speech recognition works and will help you get started.

Basic speech recognition Any technology that converts spoken words into text or commands. is built into Windows and Macintosh operating systems The software structure for a computer or mobile device; it provides a framework for applications like word processors and address books.

Try It Yourself:

You will need a microphone, preferably with a USB connector. If you don’t usually have a microphone attached to your computer, you may need to use an audio or microphone setup program.

Windows XP:  Only on Microsoft Office.  Here's some setup information.  You may need your installation CDs.

Windows Vista and 7: Go to Start Menu-->All Programs-->Accessories-->Ease of Access-->Windows Speech Recognition. The utility will take you through a "wizard" allowing you to train the system to recognize your voice. Here's a step-by-step tutorial for speech recognition on Windows 7.

Macintosh OS X: Go to Apple Menu-->System Preferences-->Speech. Choose the Speech Recognition tab, then click the Speakable Items "On" radio button. Note that this utility only gives mouse commands; it does not let you dictate text.

Third-Party Programs

Commercial speech recognition programs are more accurate, give you more control, and improve as they learn to understand your voice. The most popular are Dragon NaturallySpeaking (Windows) and Dragon Dictate (Macintosh). These programs require fast processors and lots of memory, so check the software's specifications against the machine you want to use.

Using Speech Recognition

All speech recognition programs make errors. The important thing is to correct the errors when they occur so that the system learns the user's voice and improves its accuracy. Consistent error correction can drive accuracy up to 99%. Watching for errors, however, makes the job of dictating more demanding because the user is partially distracted from the task of creating text.

Different applications like word processing, spreadsheets, and browsers have their own vocabularies, such as "format entire line to Heading 1" and "open link in new tab". Using these vocabularies makes the job easier. You can edit and create new vocabularies to suit how you use the computer.

Wireless microphones are particularly useful for people who need to use the computer entirely hands-free without being tethered via a microphone cable.

Another option with some programs is to dictate into a compatible digital recorder and transfer the audio into the computer; the speech recognition program then analyzes the speech as if it were being delivered in real time.

Limitations

Speech recognition does not work well with all voices, particularly those that are not consistent  or those with strong accents. Users need to be careful not to strain their voices when dictating and to take regular breaks.

Users also need to be able to read sufficiently well (6th grade level) to recognize and correct errors independently. In some cases, software designed to help people with learning disabilties, such as homonym checkers, can be useful. Spell checkers will not help, because everything speech recognition puts up is a correctly spelled word.

Speech recognition may not work well with some applications. It is usually most compatible with popular word processors, browsers, etc. Websites with complicated layouts may be hard to navigate through and activate the desired links. However, if it is the best or only input option for a given user, use the vocabulary creation and editing tools to improve its overall performance.

Have a question?

Ask an Expert

Upcoming Trainings