| 摘要 |
PURPOSE:To provide the adaptive post filter which can obtain an output speech signal of high quality with a small noise by using parameter values which are not discontinuous between subframes. CONSTITUTION:A synthesized speech signal obtained by a speech decoding process and pitch gains and pitch periods of the synthesized speech signal by subframes are inputted and a noise-shaped speech signal is outputted. This adaptive post filter has a pitch gain smoothing part 21 which smooths the pitch gains, a continuity decision part 23 which decide whether or not pitch periods are continuous between two successive subframes, a pitch period smoothing part 23 which smooths pitch periods including the two subframes whose continuity is decided, and a pitch emphasizing filter 31 which emphasizes the repetition of the pitches of the synthesized speech signal by using the smoothed pitch gains and pitch periods. |
