As a long-time user of voice recognition software, I thought the article
made a fair comparison of Nat and ViaVoice. However, their speed and
correct recognition rate was very low. I was unable to find what
soundboard they were using but was able to see that they only had 32 MB
of RAM. This is not enough for continuous speech. Know that the box
says "minimum 32MB RAM, but general agreement among users of Nat and
ViaVoice is that 64 MB is a minimum for good speed and recognition. The
soundboard/sound setup is very critical to the performance of this kind
of software. Having Win 95 itself, as well as all other software on the
system set up so as to provide maximum resources to the speech engine is
also critical. On a powerful enough system (mine is a 200MMX, 64RAM,
512k cache, 4MB video memory, AWE 64 Gold soundboard, VXI Gold mic) -
all software and OS set up it such a way so that I have 97% resources
free at boot up, while using NaturallySpeaking Deluxe Edition, I can
dictate as fast as I talk with very few errors. (last 5 minute timed
"reading" test = 167 wpm with .001% errors) I would highly recommend
anyone attempting to use this software with children, particularly LD
children, to:
1) Consult with a VAR or reseller with large amounts of experience with
the proper set up of this kind of hardware/software to insure that all
is correct to begin with.
2) Set up, train, and use the software yourself for a few weeks so that
you know that everything works correctly and to give you a better
understanding of the process.
Rose
There are training tips, hardware tips, VR links available from my page
at: http://idt.net/~edrose19/page7.html
This archive was generated by hypermail 2b29 : Tue Jan 04 2000 - 12:43:30 EST