Hierarchichal language models

The invention relates to the device for disclosed matter using contexts model the hierarchic structure to synchronize the voice to conversion and method. The integrated with the contexts model hierarchic structure statistical smoothly the linguistic model. The method can a process text with multiple contexts device. Each of multiple contexts device of with multiple contexts plastic the type of the hierarchic structure. Method for the invention can also includes recognizing with the text related at least one contexts model; at least contexts model for use to a processing then user phone.