Schedule Consultation

Having ¢ense and Going Realtime...

Article from the Spring 2006 issue of NVRA's Verbatim Record,

by Bettye Keyes, CCR, CSR, CVR, RVR

Most voice writers who still dictate into a recording device instead of a computer and speech recognition (SR) system have not yet taken the plunge for two main reasons: cost and time. Yet some say that these factors are not as prohibitive as another fear that holds them back, explicit warnings from select groups of CAT (computer-aided transcription) system owners regarding accuracy and functional difficulties experienced on the road to realtime success.

This has sent a message that aspirations of obtaining realtime results cannot be met with guarantees or promises, so why invest the time or money when the idea of getting an SR system to work well enough for practical use just doesn’t seem to make any sense?

Meanwhile, not learning realtime is a major stressor to reporters who feel stagnant as they watch yet other professional peers who have demonstrated realtime success move forward.

Could it be any more ambiguous?!

Concerning cost, a complete realtime system, including a high-end laptop, speech recognition engine (SRE), CAT software, and other peripherals can average around $10,000. However, you don’t have to buy a complete CAT package right now. To get started, you can buy just the SRE for under $200.

For some voice writers with busy schedules, learning how to successfully operate a voice-enabled CAT system can sometimes take up to a year, and going back to school for thorough training is simply out of the question. But did you know you could independently learn to use an SRE at home in a single day? Practically speaking, you can begin producing realtime output within a matter of hours.

Those CAT system owners who have given warnings (to non-CAT system owners) about SR technology have done so out of a desire to prepare them for the same battles they have faced. The main concern deals with a lack of time to learn how to fully operate their CAT systems in the short term, and I agree that it can be too much to master all at once. Because this detail is oftentimes overlooked in general discussion, many aspiring professionals are left with a viewpoint that SR is not viable for realtime application and that it simply doesn’t work. Optimal learning experience will turn this perception on its head.

Interestingly, reports found on the Internet provide analytical support to claims that current versions of Dragon NaturallySpeaking and ViaVoice can perform up to 99% accuracy within two weeks, or even sooner, for a general user who speaks at an average of 150 words per minute. According to other well-documented analyses,your accuracy could realistically be around 95% at 150 words per minute straight out of the box if you dictate properly. As a voice writer whose job is to dictate properly and whose occupation focuses on transcription accuracy, you will spend more time dictating and improving the SRE’s accuracy than average users. Therefore, you’ll eventually be able to achieve the 96% accuracy between 180 to 200 words per minute that Realtime Verbatim Reporters routinely demonstrate.

I must add that CAT software far outweighs what you are able to accomplish using an SRE alone, because it includes lots of additional tools and features that allow you to perform your work more accurately and efficiently. Some of the most important advantages include digital recording, audio-synced text (especially helpful during readbacks), time stamps, formatting of text with automatic punctuation insertion, easy speaker identification, word-swap and/or conflict resolution functions, on-the-fly translations, file management, and sending realtime feeds. You can learn more about available CAT programs by viewing a list of vendors at the nvra.org website.

For those not ready to make the financial commitment to a highend laptop and CAT system and who may not have time to go back to school for retraining, here’s a 3-step plan of action to turn high-dollar costs into cents:

STEP 1: Purchase SR software.

IBM ViaVoice® Release Pro USB Edition can be ordered at www.nuance.com/viavoice/pro for $189.99.

Dragon NaturallySpeaking® Preferred can be ordered at www.nuance.com/naturallyspeaking/preferred for $199.99.

[Note: ViaVoice is the recommended choice for new SR users getting started without CAT software, because its formatting capabilities allow maximum customization of textual output; a USB sound card is also included, so you will already have one to use with your CAT software when you decide to make that purchase. Be aware that these products may or may not be compatible with the CAT program you choose to use in the future and your choice may require the purchase of a different SRE product. This is an

exploratory investment only so that you can prove for yourself SR’s effectiveness.]

STEP 2: Install the SR software on your computer and follow the on-screen instructions to get set up.

The steps are pretty cut and dried. After initial setup, you’ll read some stories so the system will analyze your voice and create a voice model. The more stories you read, the better your initial SR accuracy will be. Other need-to-know basics include adding words to the vocabulary and using the document analysis feature and correction tool to improve your accuracy. This information can be found in the user’s manual or Help menu of your SRE software.

[Note: If you intend to use a speech silencer mask, use that instead of the open microphone headset that came with the SRE package. If you do not plan to use a speech silencer mask, use the open microphone headset. If you also have a USB speech processor, use that as well by inserting the plug of your mask or headset cord into the USB speech processor’s microphone jack, then plugging the USB speech processor into any USB port on your computer (be sure to use the same port each time). If you do not have a USB speech processor, you can simply insert the mask or headset plug directly into the computer through the microphone input jack. Be aware that qualities of microphones, soundcards, and other internal and external components vary. Once a voice model has been created using a certain type of dictation input device (mask vs. headset) and/or soundcard (USB speech processor vs. computer’s internal soundcard), you must continue to use those same components with the voice model you have created or you will experience poor results. If you change any one of these components, even if it’s a newer model of the same brand, you must create a new voice model.]

STEP 3: Begin doing realtime transcription.

Dictate directly into either the SR application’s document screen or word processing software of your choice.

Then when you’re ready to “make the big $ plunge,” the process of using a CAT system will be less mystifying and a realistic next step.

It may not have made sense to jump onto the realtime bandwagon at first. But now realizing that you don’t have to spend thousands of dollars and lots of time getting started, and that SR technology really does work, what are you waiting for?

So in deciding whether or not you should go realtime...

Now that it only TAKES CENTS,

doesn’t it only MAKE SENSE?

Home / FAQs / Having ¢ense and Going Realtime...