MinFrames

The minimum number of speech audio frames required to train each component of a speaker model.

During training, HPE IDOL Speech Server creates several components (mixtures), up to a maximum specified in the NMix parameter. Each component must be trained on a minimum amount of data, specified in the MinFrames parameter. If there is not enough data available, HPE IDOL Speech Server trains fewer components and the speech model performs less well.

Type: Integer
Default: 500
Range: 50–5000
Required: No
Configuration Section: sidtrain module
Example: MinFrames=300
See also: MinFrames (action parameter)

_HP_HTML5_bannerTitle.htm