To provide a device and a method for speech recognition which improves a recognition rate and enables a user to select a desired item displayed on a screen without feeling frustrated.

This speech recognition device is so configured that a spoken object display part 11 displays a plurality of spoken objects related to one item to be selected on the screen 20a of a display device and when the result of speech recognition carried out by a speech recognition part 12 by inputting speech data corresponding to the voice that the user speaks corresponds to one of the plurality of spoken objects, a decision part 13 decides that the item to which the spoken object is related is selected. Consequently, the user who selects one object only speaks so that the speech recognition part 12 can recognize one of the plurality of displayed spoken objects.