Learning and Leading - February 2011

BUYER’Sguide

Voice Recognition Software

Company

Imagine being able to write without yping or command all of the gadgets in your life by simply talking to them.

Speech recognition technologies make this possible by transcribing speech into text and serving as an alternative method of interacting with a computer.

Early programs had vocabularies that were limited to single digit numbers and required slow and careful pronunciation, but today’s programs are much better at interpreting sound waves and applying meaning to text. Cost has also come down to a price point that schools can afford.

Two types of speech recognition are often confused. Automatic speech recognition (ASR) is the conversion of any speech into text. Call-center systems and product support lines are two examples that incorporate ASR technology today. Voice recognition (VR) is a process that involves training a program to recognize the voice of a specific person. Most programs on the consumer market fit into this category.

A speaker trains a VR program as he or she recites a paragraph of text, such as the “Rainbow Passage” (which contains all the phonemes in the English language) into a microphone. The program then attempts to correct any resulting errors. Through this initial process, the application increases its vocabulary and learns how the user pronounces words.

VR technology has improved to such a degree that students with learning disabilities can now become writers without depending on others to transcribe their speech. They can speak at a natural pace, often up to 100 words per minute, when using some software.

Students for whom handwriting or typing is difficult or impossible can also use VR, and those who have difficulty putting their thoughts onto paper may benefit too. They can use the software to voice-activate a computer’s navigation commands and dictate rather than type documents into a word processing program. Spelling is accurate, though programs don’t recognize homonyms.

The accuracy of word recognition increases as users correct errors and the program learns about an individual speaker’s language patterns. If a student has sub-standard speech, does not speak distinctly, or has a thick accent, the program’s learning time will increase. Also, mere dictation does not address other aspects of writing, such as correct grammar, punctuation, sentence structure, and the smooth flow of ideas and story lines.

Computers need sufficient memory and hard drive space to use this type of software, and ambient noise in the room can interfere with accuracy. A headset with a microphone is the ideal device for entering information, and some programs include them.

Dragon, produced by Nuance, is the best known commercial brand and has the largest number of versions for home, office, and classroom use. Other PC- and Mac-based applications are available for free, but they offer fewer features.

With time, patience, and perseverance on the part of the student and those who are supporting him or her, voice recognition software can provide tremendous benefits.