Combining Monaural Noise Reduction Algorithms and Perceptive Preprocessing for Robust Speech Recognition

Abstract

In noisy and reverberant environments the performance of automatic speech recognition systems drops below acceptable levels. It has been shown before, that using a psychoacoustical model of the peripheral auditory processing, as introduced by Dau [1], yields much higher recognition rates in noisy background situations than standard pre-processing methods [2]. To further improve the robustness of a speaker-independent digit recognition system, singlemicrophone noise reduction procedures [3] have been combined with the auditory model feature extractor and continuous Hidden Markov Models as well as Locally-recurrent Neural Networks. The study shows that the recognition rates improve significantly for certain noise conditions, while the performance for clean speech is not negatively effected.

Noise reduction for speech signals by operations on the modulation frequency spectrum

“…2 show results for the PEMO/LRNN recognizer. In order to compare the MF processing to other noise-reduction methods, the algorithm by Ephraim and Malah (9) was used with the PEMO/LRNN recognizer (10). The results are also contained in Table 1.…”