1.2

13 MacSpeech DictateChapter 4 — Training
In this chapter, we cover training MacSpeech Dictate to rec-
ognize your voice. Along the way, we will share some insight
into how MacSpeech Dictate works, as well as tips for better
recognition. We will also explain how to manage multiple
profiles.
Why Do I Need To Train MacSpeech
Dictate?
MacSpeech Dictate is speaker dependent, which means it
needs to be trained to get the best accuracy. You do this by
reading one or more training stories so it can adapt to your
unique vocal characteristics.
Luckily, training MacSpeech Dictate is as simple as reading
aloud and only takes a few minutes. The Voice Training part
of the program will guide you through the process.
Speaking Properly To MacSpeech
Dictate
It is important to relax and speak in a normal conversational
tone. There’s no need to shout or project your voice. In fact,
doing so can be counterproductive and lead to poor accuracy.
It can also lead to vocal fatigue and voice strain. So relax!
Here’s a story that illustrates the approach to take regarding
training: When we were testing the first version of iListen
(our first speech recognition product) way back in 1999, we
asked two of our colleagues to run through the initial training
process. When iListen mis-recognized a word with our first
friend, she paused, continued, and all was well. When the
same thing happened to our second friend, he paused, and
then repeated the offending word in a louder and more com-
manding voice. Needless to say, the software was not cowed
by his show of dominance. When he repeated the word in his
normal voice, iListen accepted it and moved on. MacSpeech
Dictate will respond in a similar way — so remember to speak
normally.
It is also important to speak at a normal pace. MacSpeech
Dictate is a continuous speech recognition product, so it
relies on the context of words within phrases to obtain its
accuracy. If you insert… unnatural… pauses… between…
words, accuracy will actually decrease. (This is also why you
shouldn’t look at the screen while you are dictating. Waiting
for MacSpeech Dictate to catch up with you will result in these
unnatural pauses.)
Try to speak clearly. When you talk to other people, they can
fill in the words you leave out, or compensate for words you
mispronounce or slur. Computers can’t do that.
One last thing: just as MacSpeech Dictate cannot insert words
you don’t say, it also cannot eliminate words or phrases you
say, but do not want in your documents. For example, if you
tend to pepper your speech with lots of “ums,” “uhs,” “likes,”
and “you knows” MacSpeech Dictate will dutifully type what
it thinks you said.
TIP:
After reading the first training story, MacSpeech
Dictate will be familiar with your reading voice, which
may be different from your dictation voice. For an
idea of how accurate MacSpeech Dictate is at this
point, try reading it something out of a magazine or
newspaper.
Voice Training
Once you have created a profile, and set up your microphone,
MacSpeech Dictate automatically starts Voice Training
(Figure 4-1) so you can begin personalizing your profile. You
can re-enter training at any time by choosing Voice Training
from the Tools menu.
Figure 4-1
Click the microphone icon and read the text on the panel.
Note you need to say punctuation, such as COMMA and
PERIOD (US only) or FULL STOP (all dialects).
Chapter 4: Trainin