Spectral subtraction is a method for restoration of the power or the magnitude spectrum of a signal observed in additive noise, through subtraction of an estimate of the average noise spectrum from the noisy signal spectrum. The noise spectrum is estimated, and updated, from the periods when the signal is absent and only the noise is present. The assumption is that the noise is a stationary or a slowly varying process, and that the noise spectrum does not change significantly in-between the update periods. For restoration of time-domain signals, an estimate of the instantaneous magnitude spectrum is combined with the phase of the noisy signal, and then transformed via an inverse discrete Fourier transform to the time domain. In terms of computational complexity spectral subtraction is relatively inexpensive. However, due to random variations of noise, spectral subtraction can result in negative estimates of the short-time magnitude or power spectrum. The magnitude and power spectrum are non-negative variables, and any negative estimate of these variables should be mapped into a non-negative value. This nonlinear rectification process distorts the distribution of the restored signal. The processing distortion becomes more noticeable as the signal to noise ratio decreases. In this chapter we study spectral subtraction, and the different methods of reducing and removing the processing distortions.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. Boll S.F (1979), Suppression of Acoustic Noise in Speech Using Spectral Subtraction IEEE Transactions, Vol. ASSP-27, No 2, Pages 113–20, April.Google Scholar
  2. Brouti M., Schwartz R., Makhoul J. (1979), Enhancement of Speech Corrupted by Acoustic Noise, Proc. IEEE, Int. Conf. on Acoustics, Speech and Signal Processing, ICASSP-79, Pages 208–11.Google Scholar
  3. Crozier P.M. et al (1993), The Use of Linear Prediction and Spectral Scaling For Improving Speech Enhancement, EuroSpeech-93, Pages 231–2343.Google Scholar
  4. Ephraim Y., (1992), Statistical Model Based Speech Enhancement systems, Proc. IEEE, Vol. 80, No. 10, Pages 1526–55.CrossRefGoogle Scholar
  5. Ephraim Y. and Van Trees H.L. (1993), A Signal Subspace Approach for Speech Enhancement, ICASSP-93, Pages 355–58.Google Scholar
  6. Juang B. H., Rabiner L. R. (1987), Signal Restoration by Spectral Mapping, Proc. IEEE, Int. Conf. on Acoustics, Speech and Signal Processing, ICASSP-87 Texas.Google Scholar
  7. Kobayashi T. et al (1993), Speech recognition under the nonstationary noise based on the noise hidden Markov model and spectral subtraction, EUROSPEECH-93, Pages 833–837.Google Scholar
  8. Lim J. S. (1978), Evaluations of Correlation Subtraction Method for Enhancing Speech Degraded by Additive White Noise, IEEE Trans. Acoustics, Speech and Signal Processing, Vol. ASSP-26, No. 5, Pages 471–472.CrossRefGoogle Scholar
  9. Lockwood P., Boudy J. (1992), Experiments with a Nonlinear Spectral Subtractor (NSS), Hidden Markov Models and the Projection, for Robust Speech Recognition in Car, Speech Communications, Elsevier, Pages 215–228.Google Scholar
  10. Lockwood P. et al (1992), Non-Linear Spectral Subtraction and Hidden Markov Models for Robust Speech Recognition to Car Noise Environments, ICASSP-92, Pages 265–268.Google Scholar
  11. Milner B. P. (1995), Speech Recognition in Adverse Environments, PhD Thesis, University of East Anglia, UK.Google Scholar
  12. McAulay R.J., Malpass M.L. (1980), Speech enhancement using a soft-decision noise suppresion filter, IEEE Trans. ASSP, Vol. 28, no. 2, Pages 137–145, April.CrossRefGoogle Scholar
  13. Nolazco-Flores JA, Young SJ. (1994),Adapting a HMM-based Recogniser for Noisy Speech Enhanced by Spectral Subtraction, Proc. IEEE, Int. Conf. on Acoustics, Speech and Signal Processing, ICASSP-94 Adelaide.Google Scholar
  14. Porter J.E., Boll S.F. (1984), Optimal Estimators for Spectral Restoration of Noisy Speech, ICASSP-84, Pages 18A.2.1–18A.2.4.Google Scholar
  15. O’Shaughnessy D. (1989), Enhancing Speech Degraded by Additive Noise or Interfering Speakers, IEEE Commun. Mag. Pages 46–52.Google Scholar
  16. Pollak P. et al (1993), Noise suppression system for a car, EuroSpeech-93, Pages 1073–1076.Google Scholar
  17. Sorenson H.B (1993), Robust Speaker Independent Speech Recognition Using Non-Linear Spectral Subtraction Based IMELDA, EuroSpeech-93, Pages 235–238.Google Scholar
  18. Sondhi M. M., Schmidt C. E., Rabiner R. (1981), Improving the Quality of a Noisy Speech Signal, Bell Syst. Tech. J., Vol. 60, No. 8, Pages 1847–1859.Google Scholar
  19. Van Compernolle D. (1989), Noise Adaptation in a Hidden Markov Model Speech Recognition System, Computer Speech and Language, Vol. 3, Pages 151–167.CrossRefGoogle Scholar
  20. Vaseghi S.V., Frayling-Corck R. (1993), Restoration of Archived Gramophone Records, Journal of Audio Engineering Society.Google Scholar
  21. Xie F.(1993), Speech Enhancement by Non-Linear Spectral Estimation a Unifying Approach, EuroSpeech-93, Pages 617–620.Google Scholar

Copyright information

© John Wiley & Sons Ltd. and B.G. Teubner 1996

Authors and Affiliations

  • Saeed V. Vaseghi
    • 1
  1. 1.Queen’s UniversityBelfastUK

Personalised recommendations