Chapter 2 - Hardware and Software

Digital Dictation and Voice Recognition

Digital Dictation. There is a difference between digital dictation and voice recognition. Digital dictation merely records the voice in a digital mode as opposed to on an analog tape recorder. Once in a digital mode on a hard disk, etc., you can send the voice by e-mail, import it into a computer, or transfer the file to your secretary on a networked system. It replaces the analog tape dictation systems. However, it is not ready to edit in a word processing program. In order for the digital dictation to be used in a word processing program, it must be converted using voice recognition software such as DragonDictate.

Dictation is transitioning from a tape analog system to a digital floppy or hard drive system because of greater flexibility and ease of use. Attorneys still have to speak into a recording device and secretaries will still have to listen. However, there are several advantages. First, it works well over telephone lines and cell phones from any location. Secondly, the voice dictation is generally on a hard drive, so there is no longer the chance that the tape will be mislabeled. Because the voice is captured in digital mode, listening speeds can be controlled without affecting the tone of the recorded voice. This means the transcriber can slow down a memo that has been dictated at a fast pace for any number of reasons. Also, since it is in a digital format, the speaker can go back to any part of the memo or brief and insert additional comments. Such editing is unheard of with analog tape. This also allows for members of the litigation team to add their comments to any part of the tape. Digital dictation requires a lot of storage, but since the price of hard drive storage is so inexpensive, this is not a problem. You no longer have to rewind tapes. You merely delete them off the computer after a specified period of time. The transition is also easy for secretaries.

Also, since the dictation is in a digital mode, you can send it through a voice recognition program for conversion to computer editable text. The conversion accuracy will depend upon the speaker, voice recognition program and whether the speaker has “trained” the computer for his voice, etc.


Voice recognition. With voice recognition, one can dictate a pleading or letter, dial a telephone, navigate through windows, and enter data without touching a keyboard or a mouse. Voice recognition gives you the ability to insert text or data directly into your computer applications or control the Windows command interface. As the user says a word or number the computer interprets the word or number and compares it to a dictionary of tens of thousands of words. Then, it is instantly inserted into your word processing program or other application. The same process translates speech into computer commands. All window commands, even your mouse clicks, can be by voice recognition. Instead of using your mouse to navigate your way around a desktop application, simply say the word “print” to print documents. Preconfigured program controls for programs such as Microsoft Word and WordPerfect are available for purchase, or you can create them yourself. All that is needed for speech recognition is a computer, microphone, sound card, and voice recognition software.

Since the late 70’s, we have seen rapid advancement in speech recognition technology from single word recognition to continuous speech recognition. The greatest advancements have been as a result of the increasing power and declining cost of the microprocessor.

Single or “discrete” word recognition was accurate but required one to pause after each word for at least one tenth of a second. For example, you/had/to/insert/a/short/distinct/pause/between/each/word.

The new continuous speech recognition - speaking to your computer without pausing between words - permits you to talk naturally at speeds of 120 words per minute or greater. This speed is generally considered conversational speed or normal speaking. The actual output of words per minute will decrease to format text, correct errors, and add new words. Additional accuracy will increase as phrase, concept, and context recognition becomes more sophisticated.

The initial training produces amazing results; concerted practice, regular use, and training the computer to recognize your commands may be needed, depending on the complexity of your practice or duties. For example, the term “F.2d” will need to be individually added to the computer recognition system unless it comes in a custom legal vocabulary that you purchase. For skilled typists, a keyboard may still be the fastest way to enter and correct information in a computer. However, this equation is changing rapidly as the hardware and software advancements continue to get better.

Continuous voice recognition is ideal for the legal professional who cannot type or who types slowly, staff with physical impairments, and anyone who wishes to eliminate the keyboard to input text and numbers and wishes to control the computer by voice commands. You can switch between programs, issue all sorts of commands, and essentially run your computer hands free. You should definitely consider voice recognition if your attorneys are still dictating daily correspondence and forms, if you are interested in increasing productivity, or planning a computer upgrade. Voice recognition can be used by attorneys, paralegals, clerks, and receptionists – anyone who has a need to input data into a computer. Whether you are entering data into spreadsheets, word processors, databases, or sending faxes, doing research, or surfing the net, voice recognition can assist you. It is inexpensive and should reduce support staff time in providing legal services.

Using voice recognition allows your hands and eyes to be free for other tasks. Information can be entered into the computer while the legal professional is checking through past legal memoranda, reading legal quotes from books, or reviewing your case notes. One is able to work without breaking your train of thought to use a keyboard or mouse.

There are several features to consider when deciding upon which voice recognition software to purchase:

Built-in general vocabulary – How many words does the software program contain in its general vocabulary? Most programs contain at least 20,000 words or more in their general vocabulary.

Personal vocabulary – The flexibility of building or purchasing a custom vocabulary for your specialty practice area is very important. For example, if your practice is in the personal injury, contracts, or federal tort claim area, then specific legal terms are generally unique to your practice. Words or phrases such as “negligence per se”, “loss of consortium”, or “Bivens” will have to be added to your vocabulary. The software program should allow you to enter almost an unlimited number of words to your personal vocabulary module.

Some programs, such as DigitalDictate, permit you to build your custom vocabulary by taking past pleadings, cases or other workproduct from your specialty area and having the program automatically add the words that you use on a regular basis. After adding, you can train the computer to recognize the words. This method is much faster than building your library from scratch.
More and more custom vocabularies will be available for purchase as more practitioners use voice recognition.

Initial training – After acquiring a voice recognition program, you will need to “train” the computer to recognize your voice. Prior to training, I found the voice recognition to be about 50% effective. After initial training, the results increased dramatically to about 80% in accuracy. Accuracy can increase to 95% and greater as you use the system regularly, and train the computer with your special vocabulary needs.

Dictation into Common Applications – One feature to consider is whether the program allows you to dictate directly into common applications, such as WordPerfect, Microsoft Word, e-mail programs, etc.

Customized commands and macros – With voice recognition, one can create word commands to quickly insert form letters, frequently used addresses, paragraphs, sentences, and more. For example, if you say the word “thank you”, this could trigger a form paragraph that would be immediately inserted into your document.

Other considerations:

* Do the words appear immediately on the screen?

* Can you spell words using standard alphabet letter names?

* Can other users or guests use the software?

* Does it capture the voice dictation in a separate audio file to use for later playback in order to proof your document by yourself or a secretary? Unlimited audio playback is essential if you want to delegate correction of your work product. If audio accompanies your text, then a staff person can listen to your audio to correct any text problems.

* Do you have to correct mistakes on the spot, or can you save the document to correct the mistakes later? Is the correction feature easy to use?

* How easy is it to add and train new words to your vocabulary?

* Is there a hand held digital audio recorder available for later voice recognition conversion?

* Are you proficient at dictation today, or do you hand write most of your materials? If the latter is true, you should consider a pen-computing tablet or let your secretary type the materials from your handwriting papers.

* Will a slight change in your voice dictation pattern cause you difficulty? Make sure that continuous voice recognition software is able to keep up with style of speaking and dictation.

* Do you have any physical limitations causing voice recognition will be beneficial for you (such as carpel tunnel syndrome)?

* Do you have any speech limitations, such as mumbling, that will cause problems with voice recognition?

*If you decide to try voice recognition, find a trade show or conference for a demonstration or, preferably, ask a salesman to let you try the different packages. Don’t hesitate to spend money on training since it will increase your productivity.

* Does your program integrate with your favorite word processor, e-mail applications, spreadsheet, etc? Integration means the ability to operate the recognition program from within the word processing program as opposed to cutting and pasting from another program.

* Especially important is how the program lets you correct any mistakes you make. Can you use voice recognition to go back and correct mistakes, or do you need the services of a mouse?

* Does it have a vocabulary builder? This is a process where the computer reviews your most used legal documents and then will automatically create a vocabulary based on those words.

* Does the program permit voice macros? This would allow you to use initials or a few simple words to insert boilerplate language in an answer or other pleading. For example, you could say, “take a letter” and it will automatically set up a new document with your letterhead, etc. for dictation. If so, can you edit the macro easily?

* Does it provide the reverse capability – text to speech? This provides a handy method of proofing text. Usually, you can listen to the entire document or certain paragraphs in various optional voices.

*  A legal speech vocabulary is a must.

* Is the on-line documentation and help screen sufficient?

* Is it compatible with Palm and Windows CE programs?

Pay close attention to the hardware requirements. Voice recognition performs best with the fastest microprocessor, large storage, and huge amounts of RAM.

If you decide to use voice recognition, there are a number of things that will assist with your conversion to voice recognition:

* Proper microphone adjustment and location;

* Using speech recognition software requires practice and patience. It is not plug and play, and requires an investment of time. It takes generally 30 days to become proficient;

* Explain to staff that the software is not there to replace them. Using voice recognition will free up your secretary for other tasks, such as case management, etc.

The basic cost for voice recognition software is approximately $200. Costs increase as additional features are incorporated into the software and additional specialized vocabularies are purchased. To learn more about voice recognition software, check out Dragon NaturallySpeaking (