Next: Final note: the mathematics
Up: Language models and search
Previous: Search: Word graphs
The recommendations given here apply more to research
prototype systems than real life applications:
- For tasks using small vocabularies as in voice commands,
the use of N-best method is sufficient.
- For more complex tasks with large vocabularies ,
word graphs offer many advantages. In particular,
they allow the testing of different language models
without the need to rerun a full acoustic recognition.
- For speech recognition in the context of
understanding and dialogue systems (see Chapter 13), there is an
obvious need for word graphs . Without using word graphs,
it would still be possible to do the recognition job,
but by using word graphs , it is much
simpler to achieve an efficient algorithm
for the interaction between acoustic recognition,
language modelling and understanding.
EAGLES SWLG SoftEdition, May 1997. Get the book...