An Auditory Neural Feature Extraction Method for Robust Speech Recognition

This paper proposes a neural mechanism motivated system to extract noise resistant features for robust speech recognition. We use nonnegative matrix factorization to construct two layers of auditory neurons which captures the essence of speech patterns. The responses of these neurons to speech are further processed to form an auditory neural cepstral coefficient (ANCC) representation for speech recognition. We test the robustness of ANCC feature on a 51-word corpus, with recognizers trained on clean speech in noisy conditions. Compared with MFCC, ANCC shows less performance degradation and achieves satisfactory recognition accuracies in both non-stationary noise and high noise level conditions.

1. INTRODUCTION

Speech recognizers trained in clean conditions normally perform poorly in noisy environments, due to the mismatch between training and testing data. Numerous efforts have been made to improve either the front-ends [1] [2] [3] or the back-ends [4] [5] of the recognizers for noisy environments. Certain progress have been made, but the overall performance is still not satisfactory, due to several issues. First, many methods impractically require noise information beforehand. Second, methods adapted to certain noise conditions may not generalize well to different noises or even different noise levels. And third, non-stationary noises and low signal-to-noise ratio (SNR) cases is still an open problem.