A new noise reduction method using linear prediction error filter and adaptive digital filter

A technique that uses a linear prediction error filter (LPEF) and an adaptive digital filter (ADF) to achieve noise reduction in a speech degraded by additive background noise is proposed. Since a speech signal can be represented as the stationary signal over a short interval of time, most of speech signal can be predicted by the LPEF. On the other hand, when the input signal of the LPEF is a background noise, the prediction error signal becomes white. Assuming that the background noise is generated by exciting a linear system with a white noise, then we can reconstruct the background noise from the prediction error signal by estimating the transfer function of noise generation system. This estimation is performed by the ADF which is used as system identification. Noise reduction is achieved by subtracting the reconstructed noise from the speech degraded by additive background noise.

1. Introduction

In recent years, research on methods of noise reduction in a speech degraded by additive background noise is actively being done by the use of microphone array [1], spectrum subtraction (SS) [2], etc. Imperfection can be seen in the method of the noise reduction using two microphones which can be considered as a directional microphone with a blind spot in the arrival bearing of the noise. When many noise sources exist, an increase in number of microphones cannot be avoided. It is therefore important to develop a noise reduction method which uses a single microphone, and which can cancel multiple noise sources. In the systems with only one microphone, extracting a speech from a speech degraded by additive background noise requires the use of SS method. One of the SS methods [2] improves the signal to noise ratio (SNR) at the expense of processing delay, signal distortion and musical tones that arise due to the residual noise. Moreover, SS method needs an advance estimation of noise spectrum. It means that the SS method requires voice/voiceless section detector under the practical environment.