Adaptive Enhancement of Speech Signals for Robust ASR

Behavior of the least squares filter (LeSF) is analyzed for a class of non-stationary signals that are composed of multiple sinusoids whose frequencies and the amplitudes may vary from block to block and which are embedded in white noise. Analytic expressions for the weights and the output of the LeSF are derived as a function of the block length and the signal SNR computed over the corresponding block. Recognizing that such a sinusoidal model is a valid approximation to the speech signals, we have used LeSF filter estimated on each block to enhance the speech signals embedded in white noise. ASR experiments on a connected digits task, OGI Numbers95 show that the proposed LeSF based features yield an increase in speech recognition performance in various non-stationary noise conditions when compared directly to the un-enhanced speech and noiserobust RASTA filtering technique. Besides achieving noise robustness, this filtering technique yields an enhanced speech signal as a by-product. This is particularly suitable for ASR in mobile telephony networks where the noise robust feature extraction module also performs the speech signal enhancement task without incurring additional computational load.