 Suggestion
of N-best alternatives with confidence levels for
all recognitions modes |
| Optimization
of spelled words with spelling post-processor |
| Support
for incremental partial spelling |
| Robust
performance with different types of microphones |
| Beginning-of-speech
and trailing silence detection |
Exception
dictionaries to override the phonetic transcriptions
of words generated
 by
the integrated lexical expert system |
Support
for multiple pronunciation variants including phonetic
data from NaTeq and
Teleatlas |
| Optimization
for mixed use of speaker-independent words with
speaker voice tags |
| Only
one utterance needed for speaker-dependent voice
tags |
| Fast
context switching at run-time |
| Sub-bocabulary
actibation in a context at run-time |
| Words
and rules can be added to a binary grammar at run-time |
Context
can be created at run-time by compiling BNF grammars
with the integrated
grammar |
| Multiple
grammars can be merged into one context |
| Keyword
spotting |
| NLU
capabilities ranging from basic to advanced |
| Notification
on gain and SNR(Signal-to-Noise Ratio) levels |
| Multiple
instances of the engine can run concurrently |
