Speech coding device for estimating an error of power envelopes of synthetic and input speech signals

摘要

In a speech coding device for coding an input speech (So) with an AbS (Analysis by Synthesis) system, a vocal tract prediction coefficient generating circuit (201) produces a vocal tract prediction coefficient (a) from one of the input speech signal (So) and a locally reproduced synthetic speech signal. A speech synthesizing circuit (206) produces a synthetic speech signal (Sij) by using codes stored in an excitation codebook (203) in one-to-one correspondence with indexes, and the prediction coefficient (a). A comparing circuit (207) compares the synthetic speech signal (Sij) and input speech signal (So) to thereby output an error signal (eij). A perceptual weighting circuit (208) weights the error signal (eij) to thereby output a perceptually weighted signal (wij). A codebook index selecting circuit (211) selects an optimal index (I) for the excitation codebook (203) out of at least the weighted signal (wij), and feeds the optimal index (I) to the codebook (203). A power envelope estimating circuit (210) produces power envelope signals from the synthetic speech signal and input speech signal, and compares the power envelope signals to thereby estimate an error signal (Rij) representative of a difference between the envelope signals. The selecting circuit (211) selects the optimal index (I) on the basis of the error signal (Rij) and weighted signal (wij). The device is capable of reproducing a synthetic speech faithfully matching an input original speed signal without deteriorating perceptual naturalness.