BarTalk April 2000
Volume 12, Number 2

Fast facts on voice recognition software

by David J Bilinsky

Everybody’s talkin’ at me
I don’t hear a word they’re saying
Only the echoes of my mind

Words and Music by Fred Neil
Recorded by Harry Nilsson

You come into the office, turn on your computer and launch your voice recognition program. You begin your day by dictating a reply to a letter that you received that morning, saving it to your hard drive, printing the file, and then move on to review your email. Without touching your keyboard, you dictate a reply to your email, send the email, and move into your calendaring program. As you work through the day your hands barely touch your keyboard. You are producing documents, sending email, and generating work without the necessity of your secretary. Welcome to the world of voice recognition.

• Has voice recognition moved beyond the realms of science fiction and onto your desktop? Two years ago the answer was probably no. Today however, the answer is “probably”. Substantial progress has occurred in the area of voice recognition – for example DragonDictate has issued two entire versions of their software in the last two years. DragonDictate version 4 represents a substantial improvement over version 2.

• The leading contenders in the voice recognition field include: DragonDictate’s NaturallySpeaking (www.dragonsys.com), Lernout & Hauspie’s Voice Xpress (www.lhsl.com/default2.htm), IBM’s ViaVoice (www-4.ibm.com/software/speech) and Philips FreeSpeech (www.speech.philips.com). This article will emphasize DragonDictate’s product as it is the one with which I am most familiar.

• Is it expensive? You can start off with a basic voice recognition program for under $200 (which is fine for most uses). Full-blown professional versions are hundreds of dollars. Don’t like the idea of wearing a headset? Most voice recognition products now come with a mobile option – you purchase a digital recorder along with the software, dictate into the digital recorder and connect the recorder with your computer to transcribe your dictation. It is advised to purchase the recommended digital recorder for your specific software to ensure compatibility and high recognition factors.

• A high quality microphone and high-quality sound card in your computer are essential. I know some people who use a hand-held microphone and are happy with the results: however it is recommended that you use a headset as it maintains the proper distance to the microphone. Alternatively, you could use a digital recorder although you must be attentive to the question of holding the recorder sufficiently close enough for good recognition. One disadvantage of the mobile digital recorder is that you do not see the transcription on the screen as you go along, which denies you the opportunity to train the software to your voice at the same time as the software is training you on how to use it.

• You can use voice recognition to do more than simply dictate into your word processor – you can use voice recognition to launch applications, to dictate into your email program and to open and close Windows. The leading program, DragonDictate’s NaturallySpeaking, will work with Microsoft Word 97 or WordPerfect Office 2000, as well as with many other applications such as Microsoft’s Internet Explorer, Goldmine, MS Outlook and others.

• You can get your voice recognition program to read text back to you.

• Do you need high-powered equipment to run voice recognition? DragonDictate’s NaturallySpeaking is optimized for a variety of popular processors, including the Intel® Pentium® III processor, Intel Pentium II processor with MMX™ technology, AMD-K6®-III processor with 3DNow!™, and AMD Athlon™ processor. This article was dictated using DragonDictate’s NaturallySpeaking standard edition version 4 on a Pentium II processor at 266 MHz with 128K RAM. System requirements are stated as being: 48 MB RAM for Windows 95 and 98 (64 MB is recommended), 64 MB RAM for Windows NT 4.0 – although I would install more RAM memory than the recommended minimum as it will assist with the speed and accuracy of the software. Hard drive space: 200 MB free hard-disk space is the minimum. You will need a CD-ROM drive for installation. As for a sound card, Dragon recommends a Creative Labs® Sound Blaster® 16 or equivalent sound board supporting 16-bit recording – compatible cards are listed on Dragon’s web page.

• Can you use voice recognition on a laptop? Many writers have stated that laptops do not come with a sufficient quality sound card or alternatively, there is insufficient spacing between the components in the laptop which would cause your recognition factors to deteriorate. However, this article was dictated on a Dell latitude laptop and I found its recognition factors to be greater than those of a desktop machine. Dragon lists compatible notebooks on their web page – you can check compatibility before acquiring the product.

• How long does it take to start using the software? Initially, you must go through a training session which requires you to read a standard text into the software. Dragon has approached this aspect of the software with its usual concern for the user – offering excerpts from many books including Dave Barry in Cyberspace and Scott Adam’s Dogbert’s Management Tips. This initial training takes between one half to three-quarters of an hour. From that point onwards, you must stop and train the software to recognize new or different phrases and words to build your voice recognition files. This is where the time involvement can be large.

• The Dragon NaturallyMobile recorder holds up to 40 minutes of continuous speech in its built-in memory, and up to 80 additional minutes on available removable memory cards. You use the recorder as you would an analogue tape recorder: press “Record” and speak into the recorder – you can review your dictation by playing it back using the built-in speaker. Back at the office, attach the recorder to your PC’s serial port, select the recording, click “Transcribe”, and your dictation is transcribed. You can also use Dragon Mobile with Sony, Norcom and Olympus recorders (see Dragon’s web page for specific models).

• What’s next? Look for personal digital assistants that are voice-activated. Imagine talking to your PalmPilot rather than using the current printing recognition!

Sooner rather than later you will look around your law office and see people speaking to their computers – and they will be telling you to try this new technology. Will you hear a word that they are saying – or only the echoes of your mind?

Great Balls of Credit Department
Last issue I incorrectly credited Jerry Lee Lewis with writing the words and music to “Great Balls of Fire”. While “The Killer” did a masterful recording of this song, it was written by J Hammer and O Blackwell. You just can’t trust the Web to give accurate writing credits! Thanks to one reader for helping me set the record straight (see Letters to the Editor, p 24).

The views expressed herein are strictly those of the author and may not be shared by the author’s employer, the Law Society of BC. David J Bilinsky is the Practice Management Advisor at the Law Society of BC. Email: daveb@lsbc.org

This article originally appeared in the April 2000 issue of BarTalk and is reproduced here with permission of both the author and the Canadian Bar Association, British Columbia Branch.