Configuring the recognition service
The Nuance recognition service uses Nuance Recognizer to perform recognition and natural language interpretation of utterances, as requested by an application via the voice platform. It also performs acoustic adaptation, a self-tuning-engine feature which automatically improves recognition performance on deployed systems without human intervention. Acoustic adaptation analyzes data like recorded utterances and recognition results logged by the recognition service, and adapts the acoustic models accordingly.
Management Station displays commonly-used properties on the Settings tab. This table show those properties. (You can add other properties. See Speech product parameters.)
Note: If starting multiple instances of the Recognition service on a single host, use swirec_shmem_enabled on each instance to reduce memory requirements.
Service property |
Description |
Data type |
Default |
---|---|---|---|
Maximum number of n-best answers that can be returned. |
INT |
2 |
|
Audio formats that will be supplied to the speech detector (endpointer). |
STRING |
audio/basic;rate=8000| |
|
Audio formats that will be supplied to the Recognizer. |
STRING |
audio/basic;rate=8000| |
|
Number of licenses to check out during initialization of the speech detector. |
INT |
2 |
|
Number of licenses to check out during Recognizer initialization. |
INT |
4 |
|
Suppress pre-endpointing waveform logging (recordings of speech from callers). |
BOOLEAN |
true |
|
Suppresses post-endpointing waveform logging (recordings of speech from callers). |
BOOLEAN |
false |
|
Size of the memory cache. |
INT |
100 |
|
Maximum size of the disk and inet caches. |
INT |
500 |
|
Keeps preloaded grammars in the memory cache. |
BOOLEAN |
false |