Noisy Speech Enhancement Based on Multiple Statistical Models
0
조회
0
다운로드
1
인용
0
참고문헌
초록
Abstract
Most speech enhancement algorithms are based on the awwumption that speech and noise are both Gaussian in the Discrete Cosine Trasform (DCT) domain. For further enhancement of noisy speech in the DCT domain, we consider multiple statistical distrivutios (i.e., Faussian, Laplacian and Gamma) as a set of candidated to model the noise and speech. We first use the Goodness-Of-Fit 1(GOF) test in order to estavlish a set of rules to select from this set of candidates for each DCT component of noisy speech. Our evaluations illustrate that the best candidate varied from frequency to frequency bins depending on the Singal-to-Noise-Ratio (SNR) and the Power Spectral Flatness Measure(PAFM). In particular, the PSFM exhibits a strong relation with the best fit; therefore, we employ a simple recursive estimation of the PSFM in the model selection. The proposed speech engancement algorithm employs a soft earimate of the speech avsence probability(SAP) separately for each frequency bin according to the selected distribution. Both objective and subjective tests are performed for the evaluation of the proposed algorithms on a large speech database, for various SNR values and types of background noise. Our evaluations show that the proposed soft decision scheme based on multiple statistical modeling or the PSFM provides further enhancement compared with recent methods with a comparably low computational cost.
참고문헌 (27)
- "Suppression of Acoustic Noise in Speech using Spectral Subtraction", 27, pp.113~120 (1979)
- "Description and generation of spherically invariant speech-model signal", 12(2), pp.119~141
- "Speech enhancement : new approaches to soft decision", 27, pp.1231~1240 (2001)
- "Speech enhancement using warped discrete cosine transform", pp.175~177
- "Voice activity detection based on complex Laplacian model", 39(7), pp.632~634 (2003)
- "Elimination of musical noise phenomenon with the Ephraim and Malah noise suppressor", 2(2), pp.345~349
- "Speech enhancement for nonstationary noise environments", 81, pp.2403~2418
- "Optimal speech enhancement under signal presence uncertainty using log-spectral amplitude estimator", 9(4), pp.113~116
- "Speech enhancement using a minimum mean-square error short-time spectral amplitude estimator", 32, pp.1109~1121
- "Speech enhancement employing Laplacian-Gaussian mixture", 13(5) (2005)
- "Speech probability distribution", 10(7), pp.204~207
- "Probability distribution of speech signal spectral envelope" (2004)
- "Speech enhancement employing Laplacian-Gaussian mixture", 13(5), pp.896~904 (2005)
- "Order statistics in goodness-of-fit testing", 20(2), pp.209~213 (2001)
- "A spectral-flatness measure for studying the autocorrelation method of linear prediction of speech analysis", assp-22, pp.207~217
- "Subjective quality test plan for the ITU-T 4-kbit", s speech coding algorithm, jul.1999., ~
- "Spectral enhancement based on global soft decision", 7(5), pp.108~110 (2000)
- "Speech enhancement using MMSE short time spectral estimation with gamma distributed speech priors", pp.253~256 (2002)
- "Perceptual Kalman filtering for speech enhancement in colored noise", 1, pp.717~720 (2004)
- "Speech enhancement using a soft-decision noise suppression filter", 28, pp.137~145, Speech enhancement using a soft-decision noise suppression filter
- "Distributions of the two dimensional DCT coefficients for images", com-31(6), pp.835~839
- "A voice activity detector employing soft decision based noise spectrum adaptation", pp.365~368 (1998)
- "A statistical model-based voice activity detection", 6(1), pp.1~3 (1999)
- "Noisy speech enhancement using discrete cosine transform", 24(3), pp.249~257 (1998)
- "Enhanced variable rate codec, speechservice option 3 for wideband spectrum digital systems" (1996)
- "Single channel speech enhancement based on masking properties of the human auditory system", 7(2) (1999)
- "Signal subspace speech enhancement for audible noise reduction", 1(145) (2005)