From: Lee Hetherington (ilh_at_goldilocks.lcs.mit.edu)
Date: Tue Apr 01 1997 - 15:29:33 EST
No way will that work for getting a transcript of your teacher's
lecture. No way at all, and I work in the field of speech recognition.
1. Most of the commercially available systems are speaker-dependent.
They require the specific speaker to train the system rehearsing a given
set of sentences for maybe 1/2 hour or more.
2. The acoustic environment in a lecture is much, much worse than with a
close-talking mic. There will be lots of background noise,
reverberation etc.
If you're buying a speech recognition system expecting this to work for
lectures, you will be very disappointed. You're lucky if you can get
these systems to work very well under perfect conditions. I also don't
know if a 365 even has the necessary horsepower to run a recognizer (it
might).
--
Lee Hetherington, PhD.
ilh_at_sls.lcs.mit.edu
^^^
Spoken Language Systems
This archive was generated by hypermail 2.1.3 : Thu Jan 23 2003 - 09:53:31 EST