| 摘要 |
PROBLEM TO BE SOLVED: To transmit a noise signal nearer to original sound with less information quantity. SOLUTION: A blind part 101 divides an input sound signal including noise in a prescribed time unit (frame), performs a blind processing by using a Hanning window and outputs the input sound signal after the blind processing to an FFT part 102. The FET part 102 performs orthogonal conversion on the sound signal in the frame unit, which is outputted from the blind part 101, and converts the sound signal into a frequency region. A noise base estimation part 103 estimates a noise base of a signal including only noise from a sound spectrum signal in the frame unit. A thinning processing part 104 extracts amplitude information on the noise base at a prescribed frequency interval and outputs amplitude information on the extracted noise base to an amplitude quantizing part 105. The amplitude quantizing part 105 quantizes an amplitude value of the extracted noise base and outputs the result as noise amplitude information. COPYRIGHT: (C)2004,JPO
|
