Store speech, select vocabulary to recognize word
First Claim
1. A method for operating a speech recognition facility, comprising the steps of:
- receiving and storing a stream of concatenated speech items;
recognizing various such received speech items;
and associating a representation of recognized speech items for operative outputting;
said method being characterized by the following further steps;
a. from said stored speech items, selecting a first non-recognized speech item for recognizing viz à
viz an actual limited vocabulary to get a recognition representation or a failure;
b. through said failure selecting a second non-recognized speech item for recognition whilst maintaining said actual limited vocabulary and returning to a);
c. and through said recognition selecting a third non-recognized speech item, amending said actual limited vocabulary and returning to a);
d. until attaining a sufficient amount of recognitions for said operative outputting.
1 Assignment
0 Petitions
Accused Products
Abstract
For operating a speech recognition facility, the following steps are executed: a stream of concatenated speech items is received and stored; various such received speech items are recognized; and a representation of recognized speech items is associated for operative outputting. In particular, the following further steps are executed:
a. from the stored speech items, a first non-recognized speech item for recognizing viz à viz an actual limited vocabulary is selected to get a recognition representation or a failure;
b. through the failure a second non-recognized speech item is selected for recognition whilst maintaining the actual limited vocabulary and returning to a);
c. and through the recognition a third non-recognized speech item is selected, amending the actual limited vocabulary and returning to a);
d. until attaining a sufficient amount of recognitions for operative outputting.
235 Citations
8 Claims
-
1. A method for operating a speech recognition facility, comprising the steps of:
-
receiving and storing a stream of concatenated speech items;
recognizing various such received speech items;
and associating a representation of recognized speech items for operative outputting;
said method being characterized by the following further steps;
a. from said stored speech items, selecting a first non-recognized speech item for recognizing viz à
viz an actual limited vocabulary to get a recognition representation or a failure;
b. through said failure selecting a second non-recognized speech item for recognition whilst maintaining said actual limited vocabulary and returning to a);
c. and through said recognition selecting a third non-recognized speech item, amending said actual limited vocabulary and returning to a);
d. until attaining a sufficient amount of recognitions for said operative outputting. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8)
-
Specification