Exemplar-based joint channel and noise compensation

Abstract

In this paper two models for channel estimation in exemplar-based noise robust speech recognition are proposed. Building on a compositional model that models noisy speech and a combination of noise and speech atoms, the first model iteratively estimates a filter to best compensate the mismatch with the observed noisy speech. The second model estimates separate filters for the noise and speech atoms. We show that both models enable noise-robust ASR even if the channel characteristics of the noisy speech do not match those of the exemplars in the dictionary. Moreover, the second model, which is able to estimate separate filters for speech and noise, is shown to be robust even in the presence of bandwidth-limited sources.

Feature enhancement using sparse reference and estimated soft-mask exemplar-pairs for noisy speech recognition

“…This shows that the sparse mask reconstruction step is essential in enhancing ASR performance, and it can potentially improve with more accurate noise estimation. We are aware that supplementing the dictionary with artificial noise exemplars or noise exemplars extracted from the initial frames of test utterance [24], and channel compensation [25] can improve the performance of MaskMel-dict and KL-Mask-Mel-dict on mismatched noise and channel conditions. However, in this study, we are evaluating the algorithms' performance on exemplars extracted solely from the training data, without performing explicit channel compensation.…”