The amadaptadddata
module and its parameters are deprecated in IDOL Speech Server version 11.5 and later. Micro Focus recommends that you use the newer Deep Neural Network (DNN) techniques, rather than GMM-based acoustic models.
This module is still available for existing implementations, but it might be incompatible with new functionality. The module might be deleted in future.
The threshold for identifying words in adaptation data as junk.
The AmTrain
task compares each audio file with its user-supplied transcript and attempts to align each word it recognizes in the audio with its counterpart in the transcript. IDOL Speech Server produces a score that represents how closely the recognized word resembles the word in the transcript. If the words match exactly, the alignment score is zero (0). The higher the score, the more differences there are between the two words. The JunkWordThreshold
parameter sets the score above which words are identified as junk. For example, if JunkWordThreshold
is set to 2
, any words with an alignment score greater than 2 are labeled as junk.
This parameter is available only if the JunkEnabled parameter is set to True
.
Type: | Integer |
Default: | 2 |
Required: | No |
Configuration Section: | amadaptadddata module |
Example: | JunkWordThreshold=2
|
See Also: | JunkThresh (action parameter) JunkEnabled |
|