An audio reproduction device and system are capable of realizing sounds having natural sound quality with no peculiarities, and provide an excellent audio-image ascending effect which is not varied due to differences among individual persons, thereby making audio images and images coincident with each other. A frequency characteristic correction portion forms a frequency characteristic such that a sound-pressure frequency characteristic of reproduced sound radiated to a viewing position from speakers has a first peak and a second peak, wherein the center frequency of the first peak falls within the range of 6 kHz±15%, and the center frequency of the second peak falls within the range of 13 kHz±20%.