Features based on filtering and spectral peaks in autocorrelation domain for robust speech recognition

Robust Feature Extraction Using Autocorrelation Domain for Noisy Speech Recognition

“…The differentiation mentioned in equation (8) can be carried out in several ways, as discussed in reference [16]. The simple difference had shown the best results and therefore was used in reference [11], i.e.…”

“…This algorithm combines the use of the enhanced autocorrelation sequence of the noisy speech and the spectral peaks found from the autocorrelation sequence, as they are known to convey the most important information of the speech signal [11].…”

“…Dealing with the magnitude of the autocorrelation sequence, which is our concern in this paper, among the most successful methods, we can name Differentiated Relative Autocorrelation Sequence Spectrum (DRASS) [6], Short-time Modified Coherence (SMC) [7], One-Sided Autocorrelation LPC (OSALPC) [8], Relative Autocorrelation Sequence (RAS) [9], Autocorrelation Mel Frequency Cepstral Coefficients (AMFCC) [10] and Differentiation of Autocorrelation Sequence (DAS) [11]. Also, it has been shown that the use of spectral peaks obtained from a filtered autocorrelation sequence can lead to a good performance under noisy conditions [12,13].…”

Autocorrelation-based Methods for Noise-Robust Speech Recognition

“…In this algorithm (Farahani & Ahadi, 2005;Farahani et al, 2007), splitting the speech signal into frames and applying a pre-emphasis filter, the autocorrelation sequence of the frame signal is obtained using either an unbiased or a biased estimator, as shown in (4) and (5). A FIR filter is then applied to the noisy speech signal autocorrelation sequence.…”

“…In discrete domain, the definition of DPS can be approximated by the following equation Only 128 points of the spectrum are shown for clarity. The sample signal is one frame of phone /iy/ (Farahani et al, 2007).…”