practicePRO Technology Breakfasts

cheesestickspiquantAI and Robotics

Nov 17, 2013 (3 years and 8 months ago)

practicePRO Technology Breakfasts
Summary of Breakfast #2 – Voice Recognition

Presented by:
Barry Adams (Chown, Cairns)
& Paul Harte (Paul Harte Professional Corporation)
May 24, 2002
Summaries of other practicePRO Technology Breakfasts are available at BAR-eX Communications Inc. has created
online versions of some breakfasts. These allow you to hear the audio and see a screen capture
of the actual breakfast presentation. They are available at
practicePRO is a risk management initiative of the
Lawyers' Professional Indemnity Company (LAWPRO®)
Voice Recognition - Page 2
practicePRO Technology Breakfast
Voice Recognition
By Barry Adams of Chown Cairns and Paul Harte of Paul Harte Professional
Dan Pinnington:
Good morning everyone. Welcome to the second practicePRO
Technology Breakfast. My name is Dan Pinnington. I am Director of
practicePRO here at LAWPRO.
PracticePRO is LAWPRO's risk management initiative. Its goal is to
reduce claims by assisting lawyers to adapt to change, and to adopt risk
management strategies into their practices. Technology has greatly changed the
practice of law, and it is one of the many tools that can help prevent claims. To
assist lawyers to better understand and use technology, practicePRO is putting
on these breakfasts. Today's topic is Using Voice Recognition.
Someone once said to me, five years ago or so I think, that "voice
recognition has been about one year away for about ten years now." Until
recently I thought there remained a great deal of truth in this statement.
However, I now think voice recognition is finally here, although it isn't perfect and
it isn't for everyone.
Today we have two speakers that actively use voice recognition in their
practices, in similar and contrasting ways. They are going to share their
experiences with it, and some tips on how others can use it more effectively.
Barry Adams is a partner in the St. Catharines firm of Chown Cairns. Paul Harte
practises at Harte Barristers in Thornhill. We will start with Barry.
Barry Adams
I practice labour and employment law and have done so for 24 years. I
have been using PC’s from the outset, essentially since they first became
available. I have used or attempted to use virtually all of the Dragon Dictate
products. I’m not familiar with others, so I’ll confine my comments to Dragon.
Let me tell you why I bother at all. I must confess to have very bad hand
writing and limited keyboarding skills. I do use the keyboard, but not efficiently.
In addition to that, my assistant had bilateral carpal tunnel surgery which wasn’t
successful, and would prefer to do a little less keyboarding. I generate a fair
amount of paper, as we all do.
I first saw Dragon Dictate 1.0 demonstrated about 11 years ago, on a 386,
25 MHZ machine, and thought it was wonderful. I showed it to my partners who
showed much less interest in it than I did. As the years have gone by, I’ve
Voice Recognition - Page 3
practicePRO Technology Breakfast
become less messianic about the product, but have found it much more useful,
and I now use it daily. Although the sales people do overhype it, I have found it
will not change your life in any meaningful way.
As far as the nuts and bolts, let me say this about the hardware I’m using.
In the office, I have a desktop Pentium III machine, 866 mhz with 512 Megs of
RAM. We have a Novell network, and are using Windows 95 on desktops. At
home, I’ve recently upgraded to a PIII 2 Gigahertz machine with 256 Meg of RAM
running on Windows XP.
One technical issue, this product works best if you can get a lot of data
transfer between the RAM and CPU, so newer computers support the product
considerably better. I also use a digital recorder. Currently, I am using the most
recent version of DragonNaturallySpeaking version 6 (, which
appears to have been designed to operate with XP. It’s actually quite a good
product in my view.
One of the resistances of the use of this technology is that is not as
accurate as people would like it, and that is absolutely true. The problem is that
people have expected the technology to meet the sales hype. If people are
thinking about the reduction or elimination of staff, or if they are thinking about
producing documents in their final form, then they are going to be quite
It comes out of the box about 95% accurate, and with training can be
about 98% accurate. However, 98% accurate is simply not accurate enough for
lawyers. If you are dictating 250 words per page, then 2% is a lot of corrections if
you are looking to have a perfect final document.
The key point I want to make today is that the primary use that I make of
this product is for documents that do not need to be in final form, but rather are
documents that are for my eyes only. For example, if I have a client in for an
initial intake, I’ll do a memo to file. I could dictate this and have my assistant type
it up, but with the limited staffing I’ve got access to at my firm, I’d rather have my
assistant doing other things. Doing raw input is not the best use of her time. I
may take notes during the interview, but after, I’ll dictate a lengthy memo to file.
In this case I don’t have to worry about recognition errors, because if I’ve got
98% of it down, it’ll be fine. It is very useful to me when used in this way.
It is also very useful for preparing checklists, and preparing for hearings. I
find I prefer to dictate questions for witnesses. These don’t have to be letter
perfect, and they don’t even have to get the names accurate. It recognizes the
terms plaintiff and defendant quite well. In doing draft opinion letters, I also find it
very helpful, as well as draft submissions, and draft pleadings.
We use Legal Vision for our accounting purposes. I input text of dockets
by voice. If you simply send a client an account that says telephone conversation
as of such a date, it is not as useful and providing an account that includes what
the telephone conversation was about. I find this to be quite a useful use of the
I also use it if I’m reading and summarizing case law. It will recognize
citations quite well, and at the end I’ll have a useful summary for my purposes.
Voice Recognition - Page 4
practicePRO Technology Breakfast
In terms of documents that need to be prettied up, my assistant will
produce shells of letters for correspondence complete with Re: lines, the
addresses, and the signature lines. I will dictate the text of the letter, send it to
her via email, and she will block, copy and paste into the shell of the letter.
There is a downside of using the program, especially if you are producing
a document that is in final form. Although you will never have to do a spell check,
the sense of the words used may be wrong. It will put a properly spelled word in
for everything it thinks it hears, and sometimes it will not be the word that is
meant to be there. These incorrect words will not be highlighted by the
spellchecker, and they may not be caught by the grammar checker. Thus you
have to be quite diligent in proofing the document.
I mentioned that I use a voice recorder, which allows me to work without
being tethered to my computer. I also find that speaking into a mic on this
dictation like unit is easier. Although with a handheld unit there is less
recognition accuracy, because frequent changes the in position or placement of
the microphone prevent the voice file from developing. The consistency of where
the microphone is placed determines the accuracy of the product. Even a ¼ inch
can make a difference.
One other thing that I think is important about using this technology is to
properly deal with degradation issues. You will be likely using this product on
Windows. Windows has a constant problem with what is called memory leakage.
Performance will degrade over the day. To prevent this from happening I
suggest you shutdown and restart a few times over the course of the day.
Note that there are all kinds of special commands available. Avoid the
pretty up and formatting commands. It is not a good use of the technology to say
“print that”. It is simple easier and faster to press the “print” key. Also, many of
the other basic keyboard shortcuts are probably faster.
I use the product to input raw text. I spend 80% of my time trying to
enhance the product, for about 20% of its functionality. About 80% of the
functionality is text input. You can train the product to recognize exotic words,
including client names. I find that’s not a good use of my time. I also don’t have
a lot of time to develop macros. I put an asterisk in document where client name
goes, and then do search and replace to put in client name.
I recommend that you get the fastest computer you can afford; with the
maximum RAM, 256 Meg at a minimum, and ideally 512 Meg. Consider
bypassing soundcard completely by using a USB microphone. This improves
accuracy. Use a noise-canceling microphone to reduce background noise.
Remember that a consistent position of the microphone is critical. I have found
that if I put it a little further than they recommend it doesn’t tend to pick up the
breathing sounds.
The bottom line is that the cost of the hardware has come down so much
and improved so much, and the cost of the product has come down so much
that, while it won’t change your life, it will say you a whole lot of keystrokes, which
is the whole idea. It really is a no-brainer to use.
Voice Recognition - Page 5
practicePRO Technology Breakfast
Paul Harte:
I agree with almost everything Barry says. You can’t look at this as some
sort of panacea. If you’re in a secretarial/dictation pool now, this product is not
going to make it go away. But it is a great tool in certain applications.
In my practice, I have no support staff. I’ve use technology to completely
eliminate my support staff. I cannot write legibly, and I couldn’t type. In my view,
this product had to be pretty fast for it to be worthwhile. Earlier versions were
slow and had to be trained. I threw the technology on my laptop last night and
haven’t done anything with it, so you’ll get an idea of what it’s like coming right
out of the box.
I agree with Barry in that I don’t mess around with many of the extras in
the program, but I do use “scratch that” which allows you to go back and get rid
of what you just said. Otherwise I do it myself. I go back and block and correct
or dictate again. Version 6 will now self-recognize corrections, but you must
highlight whole word for this to happen. Doing this is key to improving the
accuracy of the product.
I have 2 Gig desktop with 512 RAM. You can’t have enough RAM, and
additional RAM makes it faster. There is still is a bit of delay, which can be
annoying, especially if you’re dictating something short like an email.
If I have to write a letter, I’ve created advanced macros to create the letter,
and automatically docket the time I spent on the task.
(Paul gave a demonstration of product, including drafting a letter and
entering a journal entry in Outlook which was automatically docketed - see
I spend most of my time on the phone, and use MS Outlook to make a
journal entry for every call. I use ISIS to index all of my info in Outlook. This
makes indexing and finding things very fast.
I also switch between the phone and computer. So, if I’m on the phone
with a client, I can actually dictate as I’m listening. The switching over is silent so
the client doesn’t know. It was $300 for the switch. The downside is that I wear
a headset all the time (a VXI Parrot). It’s bulky and has to be relatively close to
my mouth otherwise it’s faint, but the voice recognition is the best I’ve seen in
having used various other headsets.
You need to take a pause before you say “new line” otherwise it’s pretty
well with a natural speaking pace. I have no problem dictating very long letters,
but can type shorter letters almost as quickly with out it. I find it especially helpful
for doing journal entries. Also, for lengthy drafting, for example factums, it is very
One of my problems, because don’t have support staff and haven’t
dictated for so long, is that I don’t dictate very well. There is a certain frame of
mind for dictating. The product will work better for you if you have good dictation
During installation it integrates into almost everything other program on
your computer. This means that you can use you voice with many other
programs. It doesn’t matter what word processor you use (Word or
Voice Recognition - Page 6
practicePRO Technology Breakfast
WordPerfect), although works better with Word in terms of learning from the
corrections you make. It works well in Excel, and also in Outlook e-mails without
any difficulty. You can say “click send” to send an email.
If it types in the wrong word, and you say “correct that”, it will come up with
several options to change it to, and usually one of options is the correct one. You
can select the correct option with a voice command.
Keep the microphone in the same spot, and you have to set volume level
on microphone. If you move the microphone back and forth, even a bit, you will
get very erratic results.
I recommend getting the latest version, I use “Version 6 Preferred”. The
one thing that was annoying is that I couldn’t export import my older voice file
because I got the cheap version. I think the Preferred version was about $600.
It actually stores the voice file with the document, and it can play back your voice
as it moves over the words in the document.
I think that's about it.
Dan:Thanks Paul. Does anyone have any questions?
Q:If someone is already a fast typist, what advantage does this program
Paul:As I said, if I’m on the phone with clients, you can hear the typing, but you
can’t hear when I’m dictating. The other thing is that sometimes it’s just
nice to take a break from typing.
Barry:It also allows for multi tasking. The new digital recorders, Olympus and
Sony, allow you to record anywhere. I dictate in the car when I am
traveling. You can also use this recorder as a digital microphone.
Q:With respect to the digital recorder, how many minutes of recording time
do they have?
Barry:It depends on how much memory you put it. I think the basic units
generally come with about 40 minutes, but you can buy memory cards to
increase that. The Sony and Olympus come with 5 hours, but you’re
paying $500 for them.
Just a comment on terms of dictating styles. It is the case that when
dictating you tend to speak slower and more deliberate. The thing with
voice recognition, even though this product is called “NaturallySpeaking”, it
is not truly natural. When speaking naturally we tend to slur and run
things together. So you do have to speak a little more deliberate voice,
and slowing down just a little improves accuracy. It will catch um’s and
ah’s, but it depends on the spacing and placing of them. If you add an um
to the end of the sentence, it will not pick it up, but if you are in a middle of
a sentence and you pause then say um, it will record it.
Voice Recognition - Page 7
practicePRO Technology Breakfast
Paul:Also, don’t talk in a monotone voice because it won’t recognize it well.
You can’t dictate the way you talk. You need to speak more deliberately.
Q:What about punctuation and context of words?
Barry:It will pick up punctuation very well. The program defaults to the most
common word. The whole program is built around statistical probability of
words. So if you say “I am going to the store too for two loaves of bread”,
it will default to the most common use of to, and looks at context, but there
may be some corrections that need to be made.
Paul:For example sometimes the program has a problem with phrase “limitation
period”. If you say “limitation period period” then you’re fine, but otherwise
you get “limitation.” about 50% of the time.
The other new feature that I used was having it go through all of my
correspondence, and in essence, build a table to figure out what I was
saying. It studied my words and speech patterns. This required no time of
mine at all because I let it run overnight, and I found it was more accurate
after doing this.
Q:Both of you use Dragon Dictate and have any of you evaluated that
against the IBM Viva Voice.
Barry: I haven’t, but have seen comparisons on the net. The recognition engines
are close, and I think that the accuracy is similar, about 97%. For every
extra percentage increase in accuracy, you’re spending a lot more time
and diminishing returns come into play. If you’re looking for the perfect
solution, you’re bound to be disappointed.
Dan: Having read many reviews and comparisons over time, the common
wisdom right now is that both can be the same in terms of their accuracy.
It largely becomes a matter of personal preference as similar features may
be presented in different ways. I think it is fair to say that in the legal
world, Dragon tends to be what the majority prefers.
Barry:Part of that may be because Dragon was there first. However, Dragon is
extremely memory hungry and likes to be started first. So if you are going
to run more than one application, you’re much better off in terms of
accuracy and speed to start Dragon first. Also, I find dictating in Dragon-
pad is easy. I then do a cut and paste, so I don’t have to operate in Word
or WordPerfect. Windows programs always compete with each other.
Paul:XP has taken care of that particular issue to a large part. I also
automatically load Dragon first. I will typically run 10-12 programs at once
Voice Recognition - Page 8
practicePRO Technology Breakfast
without difficulty. Win95, Win 98, and ME have stack limitations, so when
you start adding programs you can have collisions. With XP, there are
much less of those issues.
Q:How long to get to 98%?
Paul:Not too long, I read a 10 minute passage.
Barry:You may want to retrain from time to time. When retraining it, following the
bouncing ball, tends to help. At end of day there are so many other
variables that can cause recognition to deteriorate slightly besides the
training. For example, if you speak in a certain style and then speed up,
you can lose some accuracy. You can select speed of response vs.
accuracy. For example, if you are not concerned with accuracy, you can
see display of text very quickly. But you are not going to overall change
the end result to bother with too much.
Paul:More recently I have become disciplined in highlighting the word feature to
correct text, since you have to correct it anyway, and that has gotten me to
above 98%.
Dan: If you are using the hand-held dictation unit, one option is attaching the
recorded file to an e-mail. This allows you to be able to e-mail dictation
back to the office. That is a feature that some have found helpful.
Q:The macro in Outlook you showed us, where does the docket go.
Paul:What it does is manually creates an ASCII text file, and it just throws the
ongoing entries in the bottom. I use Timeslips and it imports it
automatically. I have a paperless office. This doesn’t mean I don’t have
paper in my office, but what it means is that I have a very high-speed
printer, a very high-speed scanner (30 pages a minute) and a shredding
company that comes in every couple of weeks. So every piece of paper I
have is on my laptop, and it’s done without a great deal of time. I went
from having 8 lateral drawers to ½ a drawer. I looked at what I do 90% of
time and figured out how to automate it. The file is docketed, the
document is filed in the right spot, and it brings up WinFax with a proper
cover sheet. I found this macro on the web, and customized it to do what I
wanted it to do.
Dan:Thanks Paul and Barry. I hope everyone found this to be an informative
session. I would be pleased to hear from any of you if you have
suggestions for other topics that would be of interest. Thank you for
Voice Recognition - Page 9
practicePRO Technology Breakfast
About the practicePRO Technology Breakfasts
Technology is here to stay and lawyers must learn to use it to their advantage.
Lawyers who fail to embrace technology may find themselves unable to meet
their clients' needs and expectations, and unable to compete in an increasingly
complex practice climate.
Recognizing this, practicePRO is conducting a series of "Technology Breakfasts".
These breakfasts will be in a variety of formats, and the topics reviewed will cover
all aspects of legal technology. Some sessions will feature two or three lawyers
providing a hands-on review of how they use a particular legal technology
product. Other sessions will offer you an opportunity to see two comparable
technology products "face-off" against each other. Still other sessions will be
largely educational in nature.
Sessions will be scheduled approximately once per month, held at LAWPRO's
office at One Dundas St. West, Suite 2200, Toronto. A continental style breakfast
will be available starting at 7:45 a.m. Sessions will start promptly at 8:00 a.m. and
finish at 8:45 a.m. The cost to attend is $15. To register, please contact Nanette
O'Connor at (416)596-4623 or 1 800 410-1013, or by email at
Summaries of these breakfast sessions will appear on for
those that are outside Toronto or otherwise unable to attend. BAR-eX
Communications Inc. has created online versions of some breakfasts. These
allow you to hear the audio and see a screen capture of the actual breakfast
presentation. They are available at for $29.95.
If you are interested in hearing about a particular product or topic we would be
pleased to hear from you. Contact Dan Pinnington, Director of practicePRO at
(416) 598-5863 or 1 800 410-1013, or by email at
About practicePRO
practicePRO is LAWPRO’s innovative risk management initiative. Its mandate is
encouraging lawyers to adopt risk management strategies into their practices,
and assisting lawyers in adapting to the changing practice climate, in particular
incorporating technology into their practices. To accomplish this practicePRO
provides lawyers with various tools and resources in both print and electronic
form. Among these tool and resources are:
The “managing” series of booklets
These booklets provide insights and checklists to help lawyers better manage the
risk associated with specific practice issues. Titles include: managing the
lawyer/client relationship; managing conflicts of interest; managing the practice of
Voice Recognition - Page 10
practicePRO Technology Breakfast
investing in clients; managing a mentoring relationship, and managing practice
The Online Coaching Centre (OCC)
The OCC is an online, self-coaching tool, comprising more than 150 modules, to
help lawyers become more productive and effective in their professional and
personal lives. Topics covered include: communicating powerfully; managing
stress; overcoming procrastination; managing practice more efficiently;
developing new business opportunities; and capitalizing on emotional
Technology resources
practicePRO helps lawyers integrate technology into their practices through:
the practicePRO Technology Roadshow; practicePRO Technology Breakfasts;
and a biweekly practicePRO Technology Tip.
Wellness resources
The practicePRO Web site provides an extensive listing of links to assessment
tools, guides and resources to help lawyers address wellness and balance
Special Reports
The Special Report on Litigation explores the increase in litigation claims, the
forces driving change in litigation practice and the types of errors that underlie
litigation claims, and provides practice management tips to help reduce exposure
to claims. The Special Report on Fraud examines the new real estate fraud,
which increasingly targets lawyers, and provides tips to help lawyers avoid being
For more information visit or call 416-596-4623.