Presently, the most frequent application of this technology in the field of health
information management is in taking dictation that is largely canned reports, e.g.,
radiology, pathology, and emergency room discharge instructions. It is basically a
technology for filling in the blanks of a standard dictation template by means of voice
actuated macros. There is some capacity for "freeform dictation," but it is slow to use
because the dictator needs to pause between each word or phrase in the dictionary
and then choose from a list of possible variations when there is ambiguity of
perception, or even meaning. In other words, the software cannot interpret continuous
speech.

When voice recognition technology first came out, it could only recognize the speech
of a particular person and it had to be trained to recognize the speech of each new
user. However, greater flexibility in the algorithms which the newest programs use has
enabled many of these systems to recognize different speakers because instead of
finding an exact match between the word or phrase and a corresponding element in
the dictionary, it finds the closest match. The newer software is good enough to allow
for some variation in speakers and accent, although this greater flexibility means that
there are more choices placed on the screen from which the dictator must select. This
technology requires the dictator to do more of the editing functions that
transcriptionists now do, as well as slow down considerably so the software knows
where each word begins and ends.

What about the problem of ambient noise and high cost of hardware? High
technology unidirectional microphones have been able to eliminate or minimize
ambient noise for some time. Several years ago, the price of the hardware was much
more prohibitive than now. But the price of computers has dropped so precipitously
in the last ten months that a fast 486 can now be had for less than $3500 instead of the
$10,000 of a year and a half ago. This takes care of the most common objections to
the practicality of voice recognition technology in our field.