Saturday, April 19, 2014
Search: 

Type by Talking--The Scoop on Voice Recognition Software



Who wants to type when they can talk? A growing number of people would rather talk. More people are finding speech recognition software useful and fun for a variety of home and office applications--you just talk, and your dictated speech appears on your computer screen.

Why go through the trouble of learning how to use voice recognition software? It can speed routine tasks, making you more productive. It can become a "secretary" when you need auxiliary office help. It also can be of great help in accommodating a disability.

I became a speech-to-text user three years ago when I broke an arm and needed to find an effective way to keep working. While voice recognition software helped me stay in the loop, I found the systems available at the time difficult to program and time-consuming to "train" in voice recognition. And the results? Uneven, inaccurate text that took even more time to rewrite.

But that was then; this is now. If you tried voice recognition software in the past and weren't dazzled, you may want to look at the latest speech dictation systems. They're the reason I'm enjoying hands-free typing even though I no longer need to accommodate an injury.

The newest software is easier to install and set up, and it's better at adapting to speech patterns and vocabulary. The result is faster transcription and a significantly lower error rate.

If you tried voice recognition software in the past and weren't dazzled, you may want to look at the latest speech dictation systems.

Today's software has improved text and graphics commands and gives you more flexibility in managing your customized commands. Still, you have to work with it frequently to get the most out of it. It's important to train your system properly, and to keep training it as you go along. For example, when you use words it doesn't know, like the name of a business, it won't recognize the word unless you command the system to spell it correctly.

How do people use voice recognition software? Many find they can write more quickly when they use the software for reports, letters, and other text material. It's also an easy way for people to answer the deluge of e-mail messages they receive each day. You can use it to cut and paste, revise text, create custom commands, enter data into spreadsheets, search the Web, start programs, and open menus.

Once you get the hang of it, it's also fun to sit back with your headset on, plop your feet up on your desk, and get work done by talking to your computer. No typing necessary.

The newest software is easier to install and set up, and it's better at adapting to speech patterns and vocabulary.

Getting started

While it sounds "magical" to sit down in front of your computer and talk to it, there are a few things to know up front. Using speech recognition software requires an investment of both time and money. If your computer is more than a couple of years old, you may need to replace it before you successfully can use dictation software. Even if you have a newer model, you may need more memory or a better sound card. And while most voice recognition software is available for both Macintosh and PC (personal computer), check that out before you plunge ahead.

Dragon NaturallySpeaking Version 7, for example, runs on Intel Pentium III 500 MHz or greater and requires 128 MB RAM for standard use and 256 MB RAM for its Professional edition, which retails for $700 (see sidebar for more on costs). Depending on the application, you also need to have 300 MB to 700 MB of free hard disk space available.

Make sure you purchase quality accessories, like headsets, microphones, and recorders (although some software systems do come with accessories). Then use the software in a quiet room so the microphone doesn't pick up noise, which can be confusing during the voice recognition process.

It's important to train your system properly, and to keep training it as you go along.

What can you expect after you've trained your system? For starters, recognize that most people talk faster than they type. On average, that means you speak about 200 words to 250 words a minute when you're describing or explaining something. But typing is another matter. Fast typists can average 65 words a minute, but many people fall into the 30 words a minute category.

If you're a slow typist, the latest voice recognition software products may be a boon to your communication skills. For example, the Dragon NaturallySpeaking Version 7 product (released in 2004) accommodates up to 160 words a minute. Still, expect to have some downtime because, after the initial training, you'll need to correct mistakes as you go along. When you do that, you'll improve the program's accuracy.

Training your computer's "ears"

Voice recognition works by comparing what you say with expected choices from its list of words and phrases. It may sound like a daunting task to train your software to type tens of thousands of words accurately--just by hearing your voice--but the more you work with it, the more accurate the system becomes.

Once you get the hang of it, it's fun to sit back with your headset on, plop your feet up on your desk, and get work done by talking to your computer--no typing necessary.

Like many other first-effort experiences, training the software to identify your voice's nuances, tones, and vocabulary range can be boring. But don't be impatient. Once you've put the effort into training the software to understand you, it will get better at differentiating what you mean.

With that said, voice recognition applications can only understand the specific language and voice they're trained for. If you say a new word or pronounce something slightly differently, you'll have to add it to the system's vocabulary on file.

Staying competitive

For several years, speech recognition software mainly was popular with people in the medical and legal professions, but today a variety of professions use it to run small offices more efficiently.

One person who uses it routinely is Madison, Wisconsin attorney Steven Schaefer. He left a law firm a few years ago where he shared secretarial services and opened his solo practice. He knew he needed to be efficient to compete with larger law firms. Schaefer, who has practiced law for 27 years, remembers when an office ran far differently than it does today.

If your computer is more than a couple of years old, you may need to replace it before you successfully can use dictation software.

"A secretary would bring in the coffee and take shorthand," he says. "Today, the office environment has changed, including the fact that the cost of staffing is eating everyone alive." So about four years ago Schaefer purchased his first voice recognition software system. Today he uses the 2004 Dragon NaturallySpeaking 7 version.

Schaefer uses Dragon frequently for letters, e-mail correspondence, and for other communications as president of a nonprofit group, Camp Randall Rowing Club, Inc. "Voice recognition is a big time-saver, even though the error rate runs about 5% to 10% on average. For example, sometimes it doesn't know the difference between 'too' and 'two,' 'the' and 'a,' and 'of' and 'or.'"

Schaefer says there are three main ways to make up for the error rate:

  1. Be very careful as you go along with your writing.
    If your computer is more than a couple of years old, you may need to replace it before you successfully can use dictation software.
  2. Always use your word processor's grammar and spell check.
  3. Print out and proofread a hard copy draft before you send your document to a friend or client.

Purchasing software

Choosing software can be a daunting task, particularly because prices range widely from less than $100 to about $700, depending on how many bells and whistles you want and what applications you're going to use. Dragon, for example, has Standard, Preferred, and Professional versions. Generally speaking, the more sophisticated the customized commands are, the higher the price.

Regardless of what model you buy, it pays to get a high-quality headphone/microphone combination so you have maximum dictation and voice control.

To many established voice response users like me, 2004 is "the year of the Dragon."

While ScanSoft's Dragon has been a player in voice recognition software for several years, in 2004 the company released Dragon NaturallySpeaking Version 7, a milestone in increased accuracy and decreased time to train the system, according to Steven Schaefer, a Madison, Wisconsin attorney and frequent voice recognition software user. He spent only about one and a half hours getting the software to recognize his voice and speech patterns.

Useful resources



NCUA Equal Housing Lender

  Home & Family FinanceŽ Resource Center
  Copyright © 1997-2013 Credit Union National Association Inc.

 
University Credit Union