Noise Estimation for Speech Enhancement by the Estimated Degree of Noise WITHOUT Voice Activity Detection

M.E. Hamid, K. Ogawa, and T. Fukabayashi (Japan)


Noise estimation, the degree of noise, speech enhancement, noise subtraction


This paper presents an improved method for noise estimation of speech utterances which are disturbed by additive noise. Here, we introduce the estimated degree of noise (DON) technique for noise estimation. One advantage of this approach is that no voice activity detector (VAD) is required. Initially, noise is estimated from the valleys of the spectrum based on the harmonic properties of noisy speech, called minima value sequence (MVS). However, the valleys of the spectrum are not pronounced enough to warrant reliable noise estimates. We therefore use the estimated DON to adjust the MVS. To estimate the degree of noise, a function is prepared as a function of parameters obtained from the autocorrelation of noisy speech. Therefore, using the technique, it is possible to estimate the state of the added noise more accurately. Based on this concept, a new method is proposed that successfully updates noise even during the speech sequences without VAD. The performance of the proposed noise estimation is good when it is integrated with the speech enhancement technique.

Important Links:

Go Back