Nynex Corporation

Richard A Haddad

Walter Y Chen

A filter for filtering a speech signal to reduce acoustic noise is disclosed. In accordance with the inventive filter, the parameters of an all-pole vocal tract model are first estimated from the noisy signal using a least mean square algorithm as if no noise were present, and then the speech signal is filtered using an approximate limiting Kalman filter constructed according to the estimated parameters.

Method and filter for enhancing a noisy speech signal

A dual mode LMS channel equalizer is disclosed. The inventive channel equalizer utilizes an LMS algorithm to both identify channel parameters and to smooth the received data signal to mitigate the effects of channel additive noise. In real time operation, the inventive equalizer first identifies the channel parameters in a training period. Thereafter, the same LMS algorithm is switched to smooth the received data signal, while intermittently, the LMS algorithm is switched back to track the slowly changing channel parameters. In comparison with the conventional LMS adaptive channel equalizer, the inventive dual mode channel equalizer achieves a significant performance improvement at little additional cost.

Dual mode LMS channel equalizer

An adaptive nonlinear data driven echo canceller is disclosed. Unlike other approaches to nonlinear echo cancellation, the inventive echo canceller treats the nonlinearity of the echo path as nonlinear noise associated with transmitting data symbols through a linear echo path. In comparison to a conventional LMS data driven linear echo canceller, the inventive echo canceller utilizes only an additional small random access memory for storing a nonlinear function for estimating the effect of the nonlinear echo path on transmitted data symbols. The same LMS algorithm is used to estimate the linear echo path and the nonlinear function. The signal processing power of a conventional linear data driven LMS echo canceller is sufficient to implement the inventive nonlinear echo canceller and the convergence speed of the nonlinear echo canceller can be less than three times that of the conventional linear LMS echo canceller.

Growing community of inventors

Tuxedo, NY, United States of America

Richard A Haddad