You can specify the FrameDupl
parameter to balance accuracy against speed for audio preprocessing and language identification DNN classification.
For audio preprocessing tasks, the default value is 2
.The default value for all other tasks is 0;
you can set the parameter to 1
or 2
to improve speed, but this might have a slight impact on accuracy.
Type: | Integer |
Default: | 2 for audio preprocessing and language identification tasks 0 for all other tasks |
Ranges: | 0.0-3 |
Required: | No |
Configuration Section: | audiopreproc module |
Example: | FrameDupl=1
|
See Also: | AppDnnBase (configuration parameter)
AppDnnBase (action parameter) AppFrameDupl (action parameter) |
|