Text To Speech Performance
The text-to-speech function of Voice Xpress is generally acceptable
although it shows certain kinds of errors that imply poor testing
procedures, errors the other text-to-speech products seem to have
worked out.
The text-to-speech function mispronounced some words. For example
it would misemphasize or misvocalise one or more syllables in
"onerous", "misrecognition", "retrying", "taskbar", "redistributable",
"Pentium III", "thereby" and surprisingly,
"Voice Xpress" the product's own name. "Combines" was mispronounced
in some circumstances but not others. When reading URLs, the program
pronounced things like "domain.com" as "domain point com" instead
of the more traditional "domain dot com". When spelling out words
it could not pronounce, the letter sounds were more slurred together
than the other products making them sometimes hard to follow.
When reading back long documents, the program would first select
the entire text shifting it in such a way the tail end of the
document was visible on the screen. But reading would actually
begin at the first word of the document. Ideally the screen should
have shown, at any given time, the words being read, in this case,
the first words of the text.
We could not get text-to-speech to read mail in the mail portion
of Netscape Communicator, and odd design oversight. Text reading
did work in other word processors.
When proofreading an entire document, we were not able to find
a satisfactory way to command the text-to-speech function to pick
up and begin reading after the last most recently corrected error
or the current cursor position. There might be a command to do
that, but we could not find it in the help wizard under the keyword
"read".
The voice accent is vaguely European. Some members of the team
thought that the sibilant sounds, such as "ds" and "ts" were a
little too hard edged and therefore irritating to listen to when
having long documents read back.
More: Introduction
More: Ongoing Training
Process...
More: Dictation
into the XpressPad Applet...
You are here>: Text
to Speech Performance...
More: Integration
With Other Applications...
