| 摘要 |
PURPOSE:To facilitate the modeling of a sound source signal using a single pulse by equipping an encoding part with a means which equalizes the phase of an input speech signal for a short time. CONSTITUTION:The phase equalizing process means 2 of the encoding part 1a performs the short-time phase equalizing process so that the phase of the input speech signal converted into a discrete sample value series is zero at a peak position on a waveform appearing at pitch period intervals and a frequency spectrum analyzing means 4 extracts parameters showing the frequency spectrum of the speech signal after the phase equalization. Then a single-pulse sound source extracting means 6 outputs the position and amplitude of pulses for compressing and expressing the sound source signal of the speech signal from the speech signal after the phase equalization and a pulse sound source encoding means 9 encodes them. The pulse sound source decoding means 16 of a decoding part 1b decodes the sound source signal with the amplitude and position of the pulses sent from the encoding part 1a and a composing filter part 18 composes the decoded speech signal by composing filter processing. |
