4.3. Generating FST script voice navigation
FST scripts are generated from FST templates and the
guide annotations series described in subsection 4.2. The
states of each guide in the voice navigation model is defined
by FST templates. Using FST templates makes it possible
to provide both guide sentences and movement to show
media information such as photos and maps. Thus, it is
possible to guide using both voice and photos. The FST
templates consist of empty tags that are filled by the system
from the guide annotation series to generate the FST scripts.
One intersection may have landmarks, whereas another may
not. This means that each intersection has different guide
annotations. Therefore, having FST templates in each navigation
state from which the appropriate one can be selected
and navigation scripts generated depending on the guide
annotations is beneficial. An example of the generation of
a voice navigation FST script is shown in Fig. 9.