back to CATS home

Knowledge Base

Voice Recognition Software

About Voice Recognition (VR) Software

This entry is a stub for discussing software that translate human speech into electronic text. The intended audiences for this article include academic technology staff, assistive technology staff, faculty, and IT administrators working in CSU campuses.

Available Products

A variety of VR software products are available in the marketplace. The table below offers a summary of currently shipping products for both Windows and Macintosh operating systems.

 Product Manufacturer
 Operating System Requirements
 Product URL License Type
 Dragon NaturallySpeaking (Standard, Preferred, Professional)
 Windows 2000, XP Commercial 
 IBM ViaVoice (Personal, Standard, Advanced, Pro USB)
 Windows (98SE, ME, 2000, XP) Commercial No longer under development
 IBM ViaVoice (Simply Dictation, Mac OS X Edition)
 Nuance Mac OS X (10.1 or later)
 Commercial No longer under development
 iListen (and iListen MX)
 MacSpeech Mac OS X (10.2.8  or later for v1.6.8, 10.3, or 10.4 for v1.7) Commercial 

Usage Considerations

The accuracy of VR software is affected by numerous factors (equipment, environment, and speech patterns).

Equipment factors can significantly impact VR software performance. There is wide variation in quality among microphones and not all microphones are optimized for use with VR software. Both the quality of the microphone used during dictation and the quality of the computer's sound system used to process the dictation are known to affect VR software performance. In most cases, the best performance will be generated by headset microphones with noise-canceling circuitry. Additional characteristics to look for when selecting a microphone include: high sensitivity, the ability to adjust its position, a lengthy cable (or a wireless model) to allow somemobility. Several manufacturers of VR software provide reviews and/or recommendations for specific microphones including Nuance ( and MacSpeech (

In certain situations, specific types of microphones may be needed. For example, USB headsets allow you to bypass lower-quality integrated or card-based sound chips that often ship with laptops or less expensive desktops. Further, while some laptops do not provide sufficient power through the analog microphone port to support VR software, USB headsets receive power from the USB port and eliminate the need for self-powered microphones. Finally, while training profiles created by VR software are designed specifically for the user, microphone, and sound system of the computer on which they were created, USB headsets allow you to transfer these profiles to other machines. 

VR software is sensitive to the presence of environmental noise (especially background speech). Care should therefore be taken to select an environment that is quiet and devoid of auditory artifacts such as the echoes that result from small, enclosed spaces with many hard surfaces (e.g. wood, metal, glass).

VR software attempts to match speakers to a 'model' speech profile that is typical for that language. Speakers who manifest mild to moderate accents or speech artifacts may still achieve satisfactory results but will likely require longer training periods. Those with pronounced accents will likely find the results unsatisfactory. In some cases, VR software provides an option for use a training model designed for accented English. For example, Dragon NaturallySpeaking offers a training profile for English as spoken in the UK, Australia, India, and Southeast Asia.

You must be logged in to save your changes.