A lattice/transversal joint (LTJ) structure for an acoustic echo canceller

The least mean square (LMS) algorithm employed in the transversal filter structure has been very popular due to its simplicity. It is, however, often inappropriate to use the LMS algorithm in the transversal filter for designing an acoustic echo canceller since the echo path is usually very long and the algorithm suffers from the slow convergence. Moreover, most echo paths change their characteristics frequently according to the room environment, and thus the algorithm and structure must be chosen carefully in order to allow tracking capability. The adaptive lattice filter is an excellent alternative to improve convergence speed since the lattice structure inherently has the orthogonalization property between the backward prediction errors, but at the expense of the excessive computational load. If the input signal to be used can be modeled as an autoregressive (AR) process (say, P-th order), the reflection coefficients of the lattice filter after stage P will approach asymptotically to zero. In this case, instead of using the full-size lattice structure, the structure may be implemented in conjunction with the transversal filter structure after the P-th stage. In this paper, we propose this new lattice/transversal joint (LTJ) structure, and investigate its performance. We have found that the convergence of this new LTJ structure is almost as fast as that of the full-size lattice filter, while it maintains the computational complexity as low as that of the full-size transversal filter