Apparatus and method for adapting language model based on document clustering

The present invention relates to a device and a method for adapting a document clustering-based language model. The device for adapting a language model according to the present invention comprises: a clustering unit to cluster an entire document corpus group to at least one cluster so that documents of similar fields can be grouped; a voice recognizing unit to perform voice recognition for an input voice to generate a voice recognition result and a word grid; a cluster selecting unit to calculate distance values to the center from both the voice recognition result and the cluster, and select at least one similar cluster based on the calculated distance values; a weighting calculating unit to calculate a weighting of a language model of at least one similar cluster based on the distance value of at least one similar cluster; a language model interpolating unit to generate an interpolation language model based on the weighting and the language model; and a rescoring unit to rescore the word grid through the interpolation language model.