LOSSY POLE-ZERO MODELING FOR SPEECH SIGNALS

This paper presents an acoustic model-based lossy pole-zero modeling f or speech signals, which overcomes the limitation in the existing loss less pole-zero model that forced the numerator part of the pole-zero t ransfer function to be symmetric, We derive the lossy pole-zero model and its transfer function by employing the wave digital filter (WDF) a daptor formulas and by converting the fixed termination value -1 to a loss factor mu(0)(c) is an element of (-1, 1). Then we discuss how to determine the reflections coefficients of the lossy pole-zero model, F or this we first employ a well-performing ARMA modeling algorithm for a pole-zero type estimation of the given speech signal and then fit th e transfer function of the lossy pole-zero model to that of the ARMA m odel under Euclidean cost function. This procedure is demonstrated by an example using the Steiglitz-McBride ARMA estimation method with a s ynthetic speech signal. The lossy pole-zero modeling yields a new filt er structure-namely, three-branch lattice structure-which consists of three lattice branches with the third branch terminated by the loss fa ctor mu(0)(c) is an element of [-1, 1] and with the three branches con nected by a three-port wave adaptor characterized by the area ratio si gma is an element of [0, 1]. The three-branch lattice structure is a g eneral filter structure which becomes tbe lossless pole-zero structure when mu(0)(c) = -1 and becomes the existing all-pole lattice structur e when sigma = 0.