| 摘要 |
PROBLEM TO BE SOLVED: To reduce a preecho, a postecho, and an artifact in gain control over audio encoding and to adapt the gain control even to a signal of a spoken voice. SOLUTION: A correction function calculation part 14b discriminates between the start position of an attack signal being a signal part including an abrupt rise of the sound of an inputted audio signal and the end position of a release signal being a signal part including an abrupt fall of the sound, calculates a target peak value desirable when the current frame to be processed is brought under gain control, and calculates a correction function for the current frame to perform the gain control. The correction function calculation part 14b detects a signal part which naturally falls with a gentle gradient characteristic of a spoken voice and a signal part which naturally rises with a gentle gradient, and reduces amplification or attenuation regarding the gain control applied to the respective detected signal parts. COPYRIGHT: (C)2003,JPO
|
