Defines allophone maps for secondpass processing in Recognizer.
|
Value |
URI. One or more paths to acoustic model files. DEFAULT: (default mapfiles used) |
|
How to set |
Shown in precedence order (see Rules of parameter precedence): In a grammar using a <meta> tag. In the application's session xml in the <speechrecog> section of <resourcehints>. In Management Station set on the Nuance recognition service. If not using Management Station, set in a Recognizer configuration file (User-nrsxx.xml). |
|
Usage |
Useful when Nuance Professional Services provides custom models. |
|
Performance |
Loading models across the network is likely to cause delays. To avoid delays, preload the files (do not use a session.xml or <meta>). Once loaded, the models remain in memory. Adding too many models will degrade performance. |
Always use these parameters together, and provide the same number of files for each:
- swirec_secondpass_model_name
- swirec_secondpass_allophone_mapfile_name
- swirec_secondpass_global_fsm_name
For a description and examples, see swirec_secondpass_model_name.
