To resolve the problem that distortion may occur due to lack of quantization bits during digital processing of a low-level audio signal because an audio signal analogically occur in many cases and has features of a wide dynamic range.
A frequency (F) and an amplitude (A) of an n-bit audio signal are estimated from an inflection point (2), and an intensity (R) of a spectrum at a specific frequency (Fd) of the audio signal is calculated (6). A filter coefficient for outputting an input as it is generated in the case that the intensity (R) of the spectrum at the specific frequency (Fd) is high, and a filter coefficient (C) of a low pass filter for removing harmonic components of the estimated frequency (F) of the audio signal is generated (3) in the case that the intensity (R) of the spectrum at the specific frequency (Fd) is low. A filter part (5) uses the generated filter coefficient (C) to perform filter processing on an (n+α)-bit audio signal resulting from bit-shifting the n-bit audio signal by α bits. When the spectrum intensity (R) is high, the (n+α)-bit audio signal is outputted as it is.
COPYRIGHT: (C)2010,JPO&INPIT
Satoshi Yamanaka
Koji Minami
Takahisa Aoyagi
Yoshihiko Mori
JP2000114976A | ||||
JP10228732A | ||||
JP9191232A | ||||
JP7073186B2 |
WO2007069369A1 |
Youichi Yamagata
Masahiko Shinohara