An apparatus includes a language model group identifier configured to identify a language model group based on determined characteristic data of a user, and a language model generator configured to generate a user-based language model by interpolating a general language model for speech recognition based on the identified language model group.