Implementation of an adaptive noise canceller on the TMS320C31-50 for non-stationary environments

The results of the implementation of an adaptive noise cancelling algorithm for speech enhancement on a DSP platform are presented and discussed. Various algorithms where implemented and compared, but time-domain methods related with recursive least squares lattice-ladder filters with a posteriori updating showed the best cancelling behaviour, although at a higher computational cost, therefore, an optimised version of these filters was implemented for the DSP card. The influence of microphonic separation is also evaluated. The best results are obtained for cardioid microphones with a separation of around 20 cm, taking advantage of their angular properties to prevent speech crosstalk in the reference microphone, thus reducing the computational cost required to acceptable limits. This result is very important, since it ensures real time performance of the algorithm in a single-DSP card (farther distances involve more filter stages, which implies a higher computational cost).