A 30-day Program to Learn Speech Software
Learning to love voice recognition software takes practice, just like developing eloquent arguments.
By Steven Rich
LAWYERS may well be the best and worst candidates for using professional speech recognition software. On the one hand, the lawyerly skill of coming up with eloquent language on one's feet lends itself well to the capabilities of modern continuous speech recognition, which excels at transcribing full spoken sentences as opposed to stop and start dictation.
On the other hand, because lawyers tend to have a pervasive sense of urgency -- and very high standards -- their expectations of this emerging technology may lead to frustration when voice recognition products don't perform to standards right away. To help manage those expectations (and highlight the benefits of speech technology), here's a 30-day program to bring your office from contemplation to actual day-to-day production of notes, memos, correspondence and pleadings.
Why 30 days? A dedicated person could easily become comfortable with speech recognition software in less than a week. However, in the pressurized world of law practice, where timelines are implacable and accuracy is all, such tools are perhaps best introduced incrementally. A 30-day plan can help even the busiest attorney and equally busy staff comfortably and profitably bring speech recognition software into a thriving practice with a minimum of transitional shock.
A brief caveat: These suggestions apply whether or not you use the services of a reseller, consultant or trainer.
What can speech recognition software do for your practice? Employed correctly, it saves time and money by producing documents directly from dictation to immediate display. Along the way, it can accelerate a process that might otherwise proceed as an interactive exchange between attorney and typist over several successive drafts. It also saves time by instantly inserting blocks of frequently used text into documents when the user issues short, spoken commands (called macros). Not least importantly, it improves the quality of life for support staff by shortening the time they spend typing.
Equally important is to consider what speech recognition software will not do. It will not take the place of a knowledgeable and proactive support staff, and it isn't "plug and play." Its benefits depend on careful considerations and an up-front time investment. Despite recent reductions in initial training time required to become proficient with speech recognition software, correction and vocabulary enhancement procedures are required during the early weeks. Speech recognition software will not get it all right all of the time or work perfectly with all applications. While most speech recognition software products work impressively with voice-enabled legal forms and templates, these are not generally included with the product.
Comparative reviews in this and other legal technology magazines provide resources for choosing a vendor. When choosing among them, you may wish to keep in mind a few key hints:
- Product features vary. Those covered in this article should be considered essential.
- Many speech recognition software products do not include the specialized legal term vocabularies that enable the benefits of using speech recognition software in the practice. General language speech recognition software products cannot be sufficiently customized to compensate for lack thereof.
- Potential users should assess speech recognition software products in their native office environment to determine if the product is a good fit. Doing so can be far more informative than even the most descriptive reviews or trade-show demonstrations.
Though it's getting better, system requirements for speech recognition, particularly memory and storage space, can be heavy. Before making a purchase or requesting a trial, be certain the PC that the software will run on conforms to (or preferably exceeds) the software's minimum system requirements. Be sure the product is designed to work with your preferred word processor.
By now, your office has checked the software requirements against the PC, added memory and a sound card if needed, and chosen a word processor-compatible speech recognition software product specialized for the legal profession.
Mobile recording units that operate the software can complement legal practices well. Initially, however, put it aside.
Following the software installation, you must "train" the software on your voice through an "enrollment" the process by which the program initially learns the user's voice. Perhaps no other part of the process is as important. An enrollment that is corrupted by improper procedures will dramatically impede future performance.
To optimize the enrollment, read prescribed text to the computer clearly, exactly as instructed to, in an environment free from background and electronic noise.
Throughout the rest of the week, spend one or two hours a day reviewing the basic techniques for maximum recognition as explained in the product manual, and applying them to working on the software's provided tutorials. Patience -- with yourself, and the program -- will pay off later. Like learning to speak in court or compose a polished brief, using speech recognition software requires practice.
Be attentive to proper microphone adjustment and position, as well as refining dictation habits -- speaking not slowly, but clearly, and in complete sentences as often as possible. At this point, it may be a good idea to contact a consultant or the software company's technical support services for help.
Practitioners whose experience with word processors is limited to directing others to use them may need to invest extra time to learn some basic functions. The best speech recognition software systems allow you control of a word processor using natural and intuitive commands, but users must know the terms for the important tasks they will be asking your word processor to perform. These include opening, closing, saving and formatting documents. Make liberal use of staff in getting comfortable with these word processing features.
Before preparing for the next week's excursion into using speech recognition software for actual legal work, users should practice using the training and correction features that enable the system to learn from its mistakes. Some legal speech recognition software includes a feature for adapting to individual writing styles. Users will maximize accuracy by scanning representative documents from the practice.
Finally, choose some blocks of text that repeatedly appear in company documents -- letterhead, addresses, salutation, citations to case law or statutes -- and enter them into the software's macro text insertion feature. This allows a user to say, for example, "hourly list" and automatically insert a schedule of the partners' hourly rates into a letter.
These groundwork-laying activities can be confined to regular but brief sessions throughout week two.
Begin the third week by focusing on microphone placement habits, microphone 'tuning' (i.e. periodic adjustment for changing background noise) and conscious dictation technique. Create notes and correspondence by speaking them. More complex documents such as pleadings should be saved for week four.
Regardless of how comprehensive a vocabulary the chosen speech recognition product comes with, you will inevitably use terms (as well as proper names) that the software does not recognize. This is an opportunity to learn the simple procedures for training the program to recognize new words.
Where appropriate, this week is also a good time to introduce speech recognition software to support staff. Some attorneys have little to gain from their own use of speech recognition software. In the words of a lawyer with a seasoned paralegal, "I scribble a few lines on the yellow pad and she produces six perfect documents. Why should I need this software?" Lawyers in such fortunate positions should be reminded that the best staff members become even more effective (and less prone to repetitive strain injury) when they adopt speech recognition software.
Once lawyers have convinced staff that speech recognition software is here not to replace them but to make their lives easier, they can decide whether to involve them as full users of the software. Support staff can either create and edit documents by voice on their own -- which makes sense for the paralegal described above -- or remain limited to the use of special software features which help them edit and finalize documents lawyers dictate. The fact that the majority of the text already exists saves support staff an enormous amount of time, enabling them to focus on other tasks.
Week four is a good time to assess overall progress in dictating work. Has there been steady improvement in recognition? Have those using the software been properly using correction and vocabulary extending features? Anyone who has been using the software properly but is not satisfied with its level of recognition accuracy should perform another enrollment, making use of refinements in technique learned since the first time. Should another enrollment not improve things, consult the product's "Help" files, a consultant or technical support.
If all is well, it's time to begin dictating more complex documents, such as motions and complaints. These will challenge users to combine formatting commands, text-insertion features, macros and other advanced features of the software, such as transcribing citations.
Experiment with watching the text appear on the screen and looking away as they dictate. For some, seeing the words appear on the screen is a distraction. Others, with a different creative process, find their thoughts flow more freely if they can see the words.
At this point, you can now try a mobile unit. A digital handheld recorder lets users dictate notes and documents while are away from their PC. The software automatically downloads and transcribes digitally recorded files on the PC.
If you are willing to invest these 30 conscientious days will be richly rewarded with increased efficiency, economy and productivity. You will have created a more direct, cost-effective link between brain and document -- literally thinking out loud. In all likelihood they will be convinced that voice recognition is a viable system that benefits everyone in the law office.
Steven Rich is legal product manager for Lernout & Hauspie Speech Products, based in Burlington, Mass.