HMM-based strategies for enhancement of speech signals embedded in nonstationary noise

Abstract

An improved hidden Markov model-based (HMMbased) speech enhancement system designed using the minimum mean square error principle is implemented and compared with a conventional spectral subtraction system. The improvements to the system are: 1) incorporation of mixture components in the HMM for noise in order to handle noise nonstationarity in a more flexible manner, 2) two efficient methods in the speech enhancement system design that make the system realtime implementable, and 3) an adaptation method to the noise type in order to accommodate a wide variety of noises expected under the enhancement system's operating environment. The results of the experiments designed to evaluate the performance of the HMM-based speech enhancement systems in comparison with spectral subtraction are reported. Three types of noise-white noise, simulated helicopter noise, and multitalker (cocktail party) noise-were used to corrupt the test speech signals. Both objective (global SNR) and subjective mean opinion score (MOS) evaluations demonstrate consistent superiority of the HMM-based enhancement systems that incorporate the innovations described in this paper over the conventional spectral subtraction method.

Application of Variational Mode Decomposition on Speech Enhancement

“…Another filter named as Ephraim Malah filtering has been proposed in [10]. HMM based method also has been applied for speech enhancement in [11]. Though different methods as given in literatures along with wavelet and DCT domain have been applied by researchers the proposed method has the significance to apply in this work.…”

Speech spectral modeling and enhancement based on autoregressive conditional heteroscedasticity models

“…The probability distributions of the speech and noise processes are estimated from long training sequences of the speech and noise samples, and then used jointly with a given fidelity criterion to derive an estimator for the speech signal. Unfortunately, the HMP-based speech enhancement relies on the types of training data [13]. It works best with the trained type of noise, but often worse with other type of noise.…”

Classical Tamil Speech Enhancement with Modified Threshold Function using Wavelets

“…Speech enhancement techniques are usefully employed as a pre-processing stage while designing robust automatic speech and speaker recognition systems, low bit rate speech coders, voice communication systems and aids for hearing impaired under noisy environment. Different approaches have been proposed for speech enhancement such as Spectral subtraction, Hidden Markov Modeling, Signal subspace methods and Wavelet-based methods [3][4][5][6][7][8][9][10][11][12]. All these approaches are based on English and other European Languages.…”