Language classifiers of Language Identification System are based on robust statistical methodology. Computer models that adjust and configure the classifier to the relevant application scenario are generated in a training environment. The best possible classification profile is achieved to suit the production environment.

A (labeled) signal recording that is representative of each scenario is used in training. In a second process, the computer model generated during training determines the quality of the classifier. This involves carrying out a classification using new (labeled) signal recordings. The classification results can then be compared with the labeling and are available for use as a confusion matrix.

SLANIE Speech classifiers are distinctive for the quality of their performance, their high recognition and their robustness in use. In virtual form, they ensure that users are provided with a standardised and open interface (CORBA).