Sound data acquisition means acquires, from a sound input device, sound data representing a sound signal. First evaluation means evaluates the sound signal represented by the sound data, by using a first criterion for detecting whether or not the sound inputted into the sound input device is a breath. Second evaluation means evaluates the sound signal represented by the sound data, by using a second criterion which is different from the first criterion. Breath detection means detects that a breath is inputted into the sound input device at least when the first evaluation means makes an affirmative evaluation, in a first period in which whether or not a breath is started to be blown into the sound input device is determined, and detects that a breath is inputted into the sound input device when at least one of the first evaluation means and the second evaluation means make an affirmative evaluation, in a second period in which whether or not a breath continues to be blown into the sound input device is determined.