Masanori Tsujikawa

Model-Basedwiener Filter for Noise Robust Speech Recognition

In this paper, we propose a new approach for noise robust speech recognition, which integrates signal-processing-based spectral enhancement and statistical-model-based compensation. The proposed method, model-based Wiener filter (MBW), takes three steps to estimate clean speech signals from noisy speech signals, which are corrupted by various kinds of additive background noise. The first step is the well-known spectral subtraction (SS). Since the SS averagely subtracts noise components, the estimated speech signals often include distortion. In the second step, the distortion caused by SS is reduced using the minimum mean square error estimation for a Gaussian mixture model representing pre-trained knowledge of speech. In the final step, the Wiener filtering is performed with the decision-directed method. Experiments are conducted using the Aurora2-J (Japanese digit string) database. The results show that the proposed method performs as well as the ETSI advanced front-end in average and the variation range of the recognition accuracy according to the kind of noise is about one third, which demonstrates the robustness of the proposed method

Extended Minimum Classification Error Training in Voice Activity Detection

Voice Activity Detection (VAD) is a fundamental part of speech processing. Combination of multiple acoustic features is an effective approach to make VAD more robust against various noise conditions. There have been proposed several feature combination methods, in which weights for feature values are optimized based on Minimum Classification Error (MCE) training. We improve these MCE-based methods by introducing a novel discriminative function for whole frames. The proposed method optimizes combination weights taking into account the ratio between false acceptance and false rejection rates as well as the effect of the use of shaping procedures such as hangover.

Sudden-noise suppression with strike-portion detection based on phase linearity for speech recognition

We propose a sudden-noise suppression method for speech recognition using a phase linearity feature for noise detection. Our investigation of sound data recorded in actual retail stores shows that short, sudden noises are dominant in such environments. We also confirm the negative effect of such noises on speech recognition performance. Our method addresses this problem by focusing on sudden noises shorter than 0.5 seconds. It takes advantage of a phase feature that has been shown effective for sudden-noise detection. To accurately detect the strike-portion, which is the initial high-power part of a sudden noise, we use two different thresholds. The detected part is replaced with clean data that immediately precedes it in order to remove the noise. Results of a recognition evaluation on speech data with superimposed sudden noises show that the proposed method is effective, especially for noises with clear strike-portions, and that it can reduce the errors caused by those noises by 30–70%.