Single channel speech enhancement using a 9 Dimensional Noise Estimation algorithm and Controlled Forward March Averaging

A post processing technique is proposed to enhance speech in a single channel system. A new noise estimation algorithm is proposed in conjunction with the Controlled Forward March Averaging (CFMA) technique to enhance speech in a single channel non-stationary noisy system. We introduce a 9-Dimensional Noise Estimation (NDNE) algorithm to the Single Channel Speech Estimation (SCSE) system, that updates the estimated noise in 9 frequency sub-bands, by averaging the noisy speech power spectrum using a time and frequency dependent smoothing factor. A signal presence probability factor is calculated by computing the ratio of the noisy speech power spectrum to its local minimum, which is computed by averaging past values of the noisy speech power spectra with a look-ahead factor. The NDNE uses a non-linear thresholding map as oppose to the conventional linear thresholding. This new algorithm produced an average 7% improvement in 0 and -2.5 dB global SNR in speech corrupted with modified Babble noise. Subjective tests confirmed these results.