Advanced search
1 file | 1.71 MB

DNN-supported speech enhancement with cepstral estimation of both excitation and envelope

Author
Organization
Abstract
In this paper, we propose and compare various techniques for the estimation of clean spectral envelopes in noisy conditions. The source-filter model of human speech production is employed in combination with a hidden Markov model and/or a deep neural network approach to estimate clean envelope-representing coefficients in the cepstral domain. The cepstral estimators for speech spectral envelope-based noise reduction are both evaluated alone and also in combination with the recently introduced cepstral excitation manipulation (CEM) technique for a priori SNR estimation in a noise reduction framework. Relative to the classical MMSE short time spectral amplitude estimator, we obtain more than 2 dB higher noise attenuation, and relative to our recent CEM technique still 0.5 dB more, in both cases maintaining the quality of the speech component and obtaining considerable SNR improvement.
Keywords
DEEP NEURAL-NETWORKS, PRIORI SNR ESTIMATION, NOISE, ENVIRONMENTS, STATISTICS, MODEL, a priori SNR, speech enhancement

Downloads

  • (...).pdf
    • full text
    • |
    • UGent only
    • |
    • PDF
    • |
    • 1.71 MB

Citation

Please use this url to cite or link to this publication:

Chicago
Elshamy, Samy, Nilesh Madhu, Wouter Tirry, and Tim Fingscheidt. 2018. “DNN-supported Speech Enhancement with Cepstral Estimation of Both Excitation and Envelope.” Ieee-acm Transactions on Audio Speech and Language Processing 26 (12): 2460–2474.
APA
Elshamy, S., Madhu, N., Tirry, W., & Fingscheidt, T. (2018). DNN-supported speech enhancement with cepstral estimation of both excitation and envelope. IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING, 26(12), 2460–2474.
Vancouver
1.
Elshamy S, Madhu N, Tirry W, Fingscheidt T. DNN-supported speech enhancement with cepstral estimation of both excitation and envelope. IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING. Piscataway: Ieee-inst Electrical Electronics Engineers Inc; 2018;26(12):2460–74.
MLA
Elshamy, Samy et al. “DNN-supported Speech Enhancement with Cepstral Estimation of Both Excitation and Envelope.” IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING 26.12 (2018): 2460–2474. Print.
@article{8590200,
  abstract     = {In this paper, we propose and compare various techniques for the estimation of clean spectral envelopes in noisy conditions. The source-filter model of human speech production is employed in combination with a hidden Markov model and/or a deep neural network approach to estimate clean envelope-representing coefficients in the cepstral domain. The cepstral estimators for speech spectral envelope-based noise reduction are both evaluated alone and also in combination with the recently introduced cepstral excitation manipulation (CEM) technique for a priori SNR estimation in a noise reduction framework. Relative to the classical MMSE short time spectral amplitude estimator, we obtain more than 2 dB higher noise attenuation, and relative to our recent CEM technique still 0.5 dB more, in both cases maintaining the quality of the speech component and obtaining considerable SNR improvement.},
  author       = {Elshamy, Samy and Madhu, Nilesh and Tirry, Wouter and Fingscheidt, Tim},
  issn         = {2329-9290},
  journal      = {IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING},
  language     = {eng},
  number       = {12},
  pages        = {2460--2474},
  publisher    = {Ieee-inst Electrical Electronics Engineers Inc},
  title        = {DNN-supported speech enhancement with cepstral estimation of both excitation and envelope},
  url          = {http://dx.doi.org/10.1109/TASLP.2018.2867947},
  volume       = {26},
  year         = {2018},
}

Altmetric
View in Altmetric
Web of Science
Times cited: