N-gram language models for Polish language. Basic concepts and applications in automatic speech recognition systems

Usage of language models in automatic speech recognition systems usually give significant quality and certainty improvement of recognition outcomes. On the other hand, wrongly chosen or trained language models can result in serious degradation not only recognition quality but also overall performance of the system. Proper selection of language material, system parameters and representation of the model itself is important task during language models construction process. This paper describes basic aspects of building, evaluating and applying language models for Polish language in automatic speech recognition systems, which are intended to be used by lawyer´s chambers, judiciary and law enforcements. Language modeling is a part of project which is still early stage of development and work is ongoing so only some basic concepts and ideas are presented in this paper.