To reduce the effect of noise and improve acoustic characteristic of regenerated sound without requiring large hardware, by providing a first coefficient updating device and a first adder which adds first and second filters' output signals and outputs the result as a controlling signal to a speaker.
A first coefficient updating device 7 updates the coefficient of a first adaptive filter 6 so that a detected signal is low. An output signal from the first adaptive filter 6 is outputted to a first adder 9. A second filter 8 has in advance characteristic reverse and equivalent to characteristic which is set in a first filter 5. An acoustic signal from an acoustic signal generator 4 is outputted to the second filter 8, and an output signal from the second filter 8 is outputted to the first adder 9. The first adder 9 adds the output signals from the first adaptive filter 6 and from the second filter 8, and outputs the result as a controlling signal to a speaker 3.
TERAI KENICHI
HASHIMOTO HIROYUKI
