User guide

Section 10: DEPARTMENTS
10-11
10.7 SETTING UP SPEECH RECOGNITION
NOTE
In order to assure proper functioning of the
Speech Recognition feature, it is recom-
mended that adequate memory be installed
in the
e
maGEN
-PC.
128 MB memory should be added to an
e
maGEN
-PC with an initial memory of 128
MB, for every four additional voice ports
installed in the system -- when upgrading
from the initial four voice ports.
Speech recognition works by training the system to recognize names
and pronunciations. It is speaker independent, so you only need to
train the system once for each name. However, in some cases ac-
cents can influence the recognition and it is recommended to train
the system specifically for those names that are not recognized with
the original accent.
There are three steps to setting up the speech recognition engine:
1. Defining all the names the system is to recognize.
2. Recording the names and adding to the pronunciation table.
3. Defining for each recognized name:
• which mailbox it corresponds to and
• by which method to open that mailbox.
See Section 10.6.1.
10.7.1 Defining names
From any of the Site Parameters screens, click the Speech Recog-
nition folder link.
The Speech Recognition screen -- shown in Figure 10-4, below -- ap-
pears.